INFORMATION TO USERS

This manuscript has been reproduced from the microfilm m aster. UMI films
the text directly from the original or copy submitted. Thus, som e th esis and
dissertation copies are in typewriter face, while others may t>e from any type of
computer printer.

The quality of th is reproduction is d e p en d e n t upon th e q u ality o f th e

copy su b m itted . Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleedthrough, substandard margins, and improper
alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript
and there are missing pages, these will t>e noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g.. maps, drawings, charts) are reproduced by

sectioning the original, beginning at the upper left-hand comer and continuing
from left to right in equal sections with small overlaps.

Photographs included in the original manuscript have been reproduced

xerographically in this copy. Higher quality 6" x 9” black a n d white
photographic prints are available for any photographs or illustrations appearing
in this copy for an additional charge. Contact UMI directly to order.

ProQuest Information and Learning

300 North Zeeb Road. Ann Arbor. Ml 48106-1346 USA
800-521-0600

UMI'
 NOTE TO USERS

This reproduction Is the best copy available.

UMI
 UNIVERSITY OF OKLAHOMA

GRADUATE COLLEGE

OPTIM AL DESIG N O F CRUDE O IL D ISTIL LA TIO N PLANTS

A DISSERTATION

SUBMITTED TO THE GRADUATE FACULTY

In partial fulfillment of the requirements for the

degree of

DOCTOR OF PH ILO SOPHY

By

SHUNCHENG J I

Norman, Oklahoma

2001
 UMI N um ber: 3029622

UMI
UMI Microform 3029622
Copyright 2002 by Bell & Howell Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.

Bell & Howell Information and Learning Company

300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
 OPTIMAL DESIGN OF CRUDE OIL DISTILLATION PLANTS

A DISSERTATION

APPROVED FOR THE

SCHOOL OF CHEMICAL ENGINEERING AND MATERIALS SCIENCE

By:

Misnel

aruk Civan

Lance Lobban

/Zcj
Pakiz&Pulat

Daniel ResSsco
c Copyright by Shuncheng Ji 2001
All Rights Reserved.
 ACKNOW LEDGEM ENTS

I would like to thank all o f the Chemical Engineering Faculty and Staff at the

University of Oklahoma for their support and friendship, especially Dr. Richard

Mallinson, who provided me partial support for one and a h alf years, and Mr. Rick

Wheeler who helped me consistently with computer trouble shootings. I appreciate my

classmates, Fernando Cancino and Jose Soto for helpful discussions.

I would like to thank Dr. Faruk Civan, Dr. Lance Lobban, Dr. Pakize Pulat and

Dr. Daniel Resasco for their participation on my dissertation committee. I will be very

grateful to Dr. Bagajewicz for his guidance and patience through the entire dissertation

work.

Finally, I would like to thank my wife, Shufang; and m y daughter, Mengmeng

(Mary) for their support, encouragement and love.

IV
 TABLE OF CONTENTS

Chapter 1

Introduction...................................................................................................................................1

1.1 Description o f the problem ..................................................................................................6

1.2 Crude oil properties............................................................................................................... 7

1.2.1 Boiling range.......................................................................................................................8

1.2.2 Gravity, API......................................................................................................................11

1.2.3 Salt Content....................................................................................................................... 11

1.3 Crude distillation products................................................................................................. 11

1.4 State of the art in the design...............................................................................................14

1.5 Topics in this dissertation.................................................................................................. 17

1.6 References............................................................................................................................ 18

Chapter 2 Rigorous Targeting Procedure for the Design o f Conventional Atmospheric

Crude Fractionation U nits........................................................................................................ 20

2.1 Overview.............................................................................................................................. 20

2.2 Introduction..........................................................................................................................20

2.3 Heat demand-supply diagram........................................................................................... 26

2.4 Pump-around circuits and heat recovery......................................................................... 30

2.5 Design procedure...............................................................................................................33

2.6 Illustration...........................................................................................................................35

2.7 Effect o f minimum temperature approach...................................................................... 53

2.8 Flexibility............................................................................................................................55

2.9 Reboiler............................................................................................................................... 55
2.10 Applicability to other crudes............................................................................................57

2.11 Conclusion......................................................................................................................... 58

2.12 Nomenclature....................................................................................................................58

2.13 References.......................................................................................................................... 60

Chapter 3 Rigorous Targeting Procedure for the Design o f Crude Fractionation with Pre­

flashing........................................................................................................................................ 63

3.1 Overview...............................................................................................................................63

3.2 Introduction.......................................................................................................................... 63

3.3 Effect o f pre-flashing on products.................................................................................... 67

3.4 Carrier effects...................................................................................................................... 71

3.5 Steam Stripping: Effect o f the temperature......................................................................80

3.6 Steam Injection....................................................................................................................83

Energy requirements for stripping agents............................................................................... 84

3.8 Effect o f pressure................................................................................................................ 85

3.9 Comparison for lower AGO yield...................................................................................87

3.10 Effect o f pre-flash tem perature...................................................................................... 90

3.11 Feed location o f pre-flash drum Vapors.........................................................................91

3.12 Conclusion......................................................................................................................... 95

3.13 Nomenclature.....................................................................................................................96

3.14 References.......................................................................................................................... 97

Chapter 4 On the Energy Efficiency of Stripping-Type Crude Distillation.......................99

4.1 Overview.............................................................................................................................. 99

4.2 Introduction.......................................................................................................................... 99

VI
4.3 Crude vaporization patterns and heat demand..............................................................103

4.4 Features o f Stripping type design.................................................................................. 108

4.5 Energy Targeting.............................................................................................................109

4.6 Comparison with the conventional design....................................................................115

4.7 Conclusions.......................................................................................................................121

4.8 References........................................................................................................................114

4.9 Appendix..........................................................................................................................122

Chapter 5 Design o f Crude Distillation Plants with Vacuum Units. I. Targeting........ 126

5.1 Overview..........................................................................................................................126

5.2 Introduction..................................................................................................................... 126

5.3 Alternatives for atmospheric-vacuum distillation........................................................130

5.4 Vacuum oil distillation................................................................................................... 134

5.5 Vacuum tower specifications......................................................................................... 140

5.6 Targeting procedures...................................................................................................... 141

5.7 Results for the light crude.............................................................................................. 144

5.8 Heavy crude..................................................................................................................... 153

5.9 Conclusion........................................................................................................................154

5.10 References...........................................................................'......................................... 157

5.11 Appendix........................................................................................................................158

Chapter 6 Design o f Crude Distillation Plants with Vacuum Units. II. HEN Design.... 161

6.1 Overview...........................................................................................................................161

6.2 Introduction...................................................................................................................... 161

6.3 Energy Targets................................................................................................................. 164

VII
6.4 Heat exchanger network.................................................................................................. 170

6.5 Stream matching pattern and its representation............................................................171

6.6 Multi-period model......................................................................................................... 178

6.7 Results and Discussion.................................................................................................... 181

6.8 Conclusion........................................................................................................................ 184

6.9 Nomenclature.................................................................................................................... 188

6.10 References....................................................................................................................... 189

Chapter 7 New flowsheets..................................................................................................... 192

7.1 Introduction....................................................................................................................... 192

7.2 Use o f the heat demand-supply diagram ....................................................................... 192

7.3 Distillation sequencing..................................................................................................... 197

7.4 Vacuum distillation.......................................................................................................... 202

7.5 References.........................................................................................................................204

vin
 ABSTRACT

Crude distillation is the first processing unit in the refinery. It separates the crude

in to distillates, which are either intermediate products further processed or raw materials

for the chemical industry. Crude distillation consumes fuel in the equivalent of 2% o f the

crude processed. This dissertation proposes a rigorous procedure that uses simulators that

are able to rigorously represent the column operating conditions as a function o f the

design parameters and complements it with heat exchanger network design methods.

Targeting procedures for both atmospheric distillations and complete distillations are

presented. These procedures use a heat demand-supply diagram and commercial software

to determine the energy targets for a plant processing several crude oils. These

procedures allow each crude to be processed at optimal conditions. Two important

alternative designs, the preflash design and stripping-type design have been studied in

detail. A rigorous definition for “carrier effect” is provided and the role of light

components in the crude distillation is analyzed. The impact o f the different heating

pattern in the stripping-type design is analyzed and compared with the conventional

design. It is concluded that that a stripping-type distillation is less competitive than the

conventional design in terms o f energy consumption and investment cost. The problem o f

designing a multi-period heat exchanger network for a complete distillation plant is

addressed. The plant has a preflash drum to facilitate the processing of light crudes. A

new concept, matching pattern, is proposed to capture the heat exchange topology

features. The dissertation also proposes a methodology to develop new designs and

provides some examples that are proven promising.

IX
 Chapter 1 Introduction

Crude distillation is the first processing unit in the refinery. It separates the crude

in to distillates, which are either intermediate products further processed or raw materials

for the chemical industry. Crude distillation consumes fuel in the equivalent of 2% o f the

crude processed. Because o f its large scale, even a small percentage of energy reduction

is economically significant. In spite of its importance, procedures for its optimal design

are rather scarce compared with other large-scale chemical processes.

A schematic diagram for crude distillation is shown in Figure 1.1. The crude oil is

split in the atmospheric tower into fuel gas, naphtha, kerosene, diesel, gas oil and

atmospheric residue (topped crude). The atmospheric residue is sent to the vacuum tower

to be further separated into light vacuum gas oil, heavy vacuum gas oil and vacuum

residue.

condenser Vacuum tower

naphtha To Ejectors

Preflash drum
steam
kerosene
VLGO
steam
Crude Oil
diesel VHGO

Steam
Vacuum Residue
sour water Furnace
Vacuum furnace
Desalter

Figure 1.1 A schematic diagram for crude distillation

 The crude distillation designs evolved in many years of practice. Figure 1.2 shows

a simplified version o f the atmospheric column. There are five products coming out from

this column: naphtha from the top, residue from the bottom, kerosene, diesel and gas oil

from the intermediate trays. The three products (kerosene, diesel and gas oil) withdrawn

from the side o f the column are also called side withdrawals.

The side withdrawals can not meet their specifications because they contain

substantial amount o f light components. The light components are removed in side

strippers (Figure 1.3).

» water
^ naphtha

kerosene

diesel

>■ gas oil

Cnide

Flash zone
steam

Residue

Figure 1.2 The amiospheric distillation column (design A)

 ^ water
naphtha

steam Stripper 3

►kerosene

steam Stripper 2
diesel

Crude steam Stripper 1

^ \
^ ---------
* gas oil
Flash zone
steam

Residue

Figure 1.3 The atmospheric distillation column with side strippers (design B)

Design B is a feasible scheme but not a good one. In this design, all heat is

removed from the condenser. This results in vapor and liquid traffic increasing

throughout the tower from the bottom to the top. The diameter o f the top section o f the

column has to be large to accommodate the high traffic (Watkins, 1979). This scheme of

heat removal is also not appropriate for heat recovery because heat from the condenser at

a low temperature level (Watkins, 1979, Bagajewicz, 1998).

Design B can be improved by removing part o f the heat somewhere below the top

tray. There are two types o f schemes (Watkins, 1979): pump-around reflux and pump-

back reflux.
 The pump-back scheme (Figure 1.4) withdraws the liquid from a tray, cool it

down and pumps it back to the column at a point several trays below the withdrawal

point. Note the amount o f heat that can be removed by the pump-back scheme is limited

by the liquid entering the withdraw tray.

water
►naphtha
Pump-back 1
‘A
steam Stripper 3
Pump-back 2
C kerosene

steam Stripper 2
Pump-back3 ^ diesel

Crude steam Stripper 1

gas oil
Flash zone
steam

Residue

Figure 1.4 Pump-back reflux

The pump-around scheme (Figure 1.5) uses an externally circulated and cooled

stream for partial heat removal. Because the liquid is recycled in the pump-around zone,

the amount of liquid withdrawn can be several times larger than the liquid reflux entering

the pump-around zone. The pump-around scheme is the design used in the refineries.
 water
naphtha
PAl
c :
steam Stripper 3
P A 2 ^
kerosene

H j steam Stripper 2
PA3
c :
Crude steam Stripper 1

gas oil
Flash zone
steam

Residue

Figure 1.5 Pump-around reflux — some heat removed from the pump-around

circuits

Thus, the conventional design using side strippers and pump-arounds (Figure 1.5)

appeared 70 years ago and still the dominant design. Because the trade-off between

energy cost and capital cost has changed profoundly in the past 70 years, it is necessary

to check whether the assumptions used to design this process are still valid and propose

an updated procedure. Existing procedures (Watkins, 1979) are not accurate in nature

because they rely on empirical charts. Another problem is that these procedures do not

take into account the impact o f heat integration and are therefore energy inefficient. In
addition, new methodologies are available for tackling certain design problems that did

not exist at the time these procedures were developed.

Another motivation for this thesis is to study some interesting design alternatives.

For example, the addition o f pre-fractionation columns to this conventional design

(Figure 1.6) is practiced sometimes. There have always been disputes as to whether these

designs could lead to smaller energy consumption. This thesis evaluates these designs in

a systematic way.

i condenser
-►

PAl heavy naphtha

I light naphtha
c :
4—Î----- »■ Steam
PA2
Crude oil cd kerosene

desalter — » r *1 Steam
stean

C=l PA3
c : diesel

sour water
A Steam

gas oil
• - 0 - steam
fumace ► residue
prefractionation
tower mam tower
Figure 1.6 Pre-fractionation Design

1.1 Description of the problem

The crude distillation problem is the following. Given:

• A set of crude oils to be processed and their properties

• A set o f specifications for products

• Maximal allowable processing temperatures o f the crudes

• Final temperatures o f products

• Available utilities, their temperatures, and their costs

determine the distillation column(s) and the associated heat exchanger network that will

minimize the annualized cost o f the equipment plus the annual cost o f utilities.

1.2 Crude oil properties

Crude oil is very complex. The compounds in crude oils can be grouped into 3

major categories: Paraffinic hydrocarbons, naphthenic hydrocarbons and aromatic

hydrocarbons. Paraffinic hydrocarbons are saturated compounds. Naphthenic

hydrocarbons are saturated cyclic hydrocarbons. Aromatic hydrocarbons contain

unsaturated cyclic compounds.

With gas chromatographic methods, more than 70 different components have

been identified from very light hydrocarbons (less than 10 carbon atoms) to very heavy

large hydrocarbon molecules. The relevant properties to crude distillation are boiling

ranges, density and salt content. Boiling ranges are used because of the large amount of

compounds which are very expensive to identify. Density helps characterize the crude

because it tells about its proportion of light hydrocarbons. Finally salt comes with the

crude and it is harmful. So it needs to be removed.

1.2.1 Boiling range

The boiling range of the crude oil provides information about the quantities of the

various products present. The commonly used distillations are true boiling point (TBP)

distillation (TBP) and ASTM distillation.

True Boiling Point (TBP) distillation is a batch distillation using a large number

of trays and a high reflux ratio. The temperature-volumetric yield curve is constructed

representing the actual boiling point o f the hydrocarbon material present at the volume

percentage point (Watkins, 1979). Most commonly the column has 15 theoretical trays

and is operated at a reflux ratio o f 5. TBP distillations are usually reported at 760 mm Hg,

even though the highest boiling portion of the mixture is distilled out at reduced pressures

(30 to 40 mm Hg). The TBP data (Figure 1.7) can be used to define the average normal

boiling points for pseudocomponents. The pseudocomponent concept is very useful for

simulations. During the TBP distillation, if the distillates are collected over a small

temperature range, the whole crude is then divided in fractions. Each fraction is still a

mixture o f m any unknown components, but the components in one fraction have a similar

boiling point. It would be convenient for distillation calculations to treat these fractions as

if they were pure components. They are called pseudocomponents.

In practice, it is not necessary to collect those narrow fractions experimentally.

The pseudocomponents can be obtained by setting up cutting points on a smooth TBP

distillation curve. Figure 1.8 shows the first three pseudocomponents. It can be seen that

a psedocomponent is defined by the interval between two adjacent cut points and its

percentage in the crude is the differences between values at adjacent cut points. The

normal boiling point (NBP) of each fraction is determined as a volume-fraction average

by integrating across the cut range. These average boiling points are used to calculate

other thermophysical properties for each pseudocomponent (PRO/II Workbook, 1996).

1000

400
LÜ
& 200
S
UJ
I-

-200
0 20 40 60 80 100
PERCENT DISTILLED

Figure 1.7 Typical TBP curve

1 00

80

60
U
UJ
ae 40

I
UJ
20
Q.
S
-20
-40 2nd 3rd P C
PC PC
-60
0 2 4 6 8 10 12
P E R C E N T DISTILLED

Figure 1.8 Definition o f pseudo-components based on TBP curve

PC: pseudo-component
 The disadvantage o f TBP distillation is that the equipment is expensive and the

test takes a long time to carry out. An easier method is ASTM D-86 distillation. ASTM

D86 distillation is carried out in a glass flask equipped with an electric heater and a water

cooler. The oil sample is heated in the flask and the resultant vapor is condensed and

collected in a receiver at laboratory pressure. The temperature versus volumetric amount

collected is recorded. Virtually no fractionation occurs in this distillation. The initial

boiling point (IBP) is defined as the temperature at which the first drop of liquid is

collected. The final boiling point or end point is defined as the maximum temperature

reached during the distillation. ASTM D86 distillation is typically used for light and

medium petroleum products and is carried out at atmospheric pressure. ASTM D l l 60

distillation (PROU manual, 1996) is similar to ASTM D86 distillation, but is used for

heavier petroleum products and is often carried out under vacuum, sometimes at absolute

pressures as low as 1 mm Hg.

One can convert these curves to TBP using correlation (Nelson, 1958, API

handbook, 1970).

Equilibrium flash vaporization (EFV) is a test used to determine the vaporization

ratios o f crude oils as a fimction o f temperature. In this test, the crude oil is continuously

heated without separating the vapor from the remaining liquid (Nelson, 1958). The

temperature versus liquid amount is recorded. To obtain the EFV curve, a series o f runs

at different temperatures are carried out, and each nm constitutes one point (of

temperature and percentage vaporized) on the flash curve.

10
1.2.2 Gravity, API

API is a density scale used in the petroleum industry. It can be calculated from

specific gravity is the following (Gray, 1994):

o A P / = 045.5/ s p g r ) - 131.5 (1)

In the above equation, specific gravity refers to the weight per unit volume as

compared to water at 60 °F. Most crude have a gravity in the range o f 20 to 45 API. 20

corresponds to heavy crudes and 45 to lighter crudes.

1.2.3 Salt Content

The salts in crude oil are primarily chlorides, sulfates and carbonates o f sodium,

calcium and magnesium. The salt may deposit on heat transfer surfaces and cause

fouling. Chloride salts m ay also decompose to form acids and cause corrosion. The salt

content of the crude oil is required to be less than 10 lb /1000 bbl for crude distillation

(Gray, 1994).

Desalting is performed by mixing the crude oil with 3 to 10 volume per cent water

at temperatures from 90 to 150 °C (200-300 °F). Both the water/oil ratio and the

temperature of operation are functions of the density o f the crude oil. Typical operating

conditions are shown in Table 1.1 (Gray, 1994).

1.3 Crude distillation products

Typical fraction cut points and boiling ranges for atmospheric and vacuum tower

fractions are given in Table 1.2.

11
 Table 1.1 Typical Operating Condition for Desalting

API Water Wash, vol % Temperature °C (°F)

API>40 3-4 115-125 (240-260)

30<APK40 4-7 125-140(260-280)

APK30 7-10 140-150 (280-300)

Table 1.2 Typical distillates and its use

Boiling ranges
Fraction ASTM (°F) TBP (°F) Procesing use
Naphtha ( heavy straight- 180-400 190-380 Reforming stock
run gasoline)
Kerosene 330-540 380-520 Kerosene, jet-50 cut

Light gas oil (LGO) 420-640 520-610 Diesel fuel

Heavy gas oil (HGO) 550-830 610-800 Catalytic cracker or
hydrocracker feed
Vacuum gas oil (VGO) 750-1050 800-1050 Deasphalter,
catalytic cracker
feed or
hydrocracker feed
Vacuum reduced crude 1000+ 1050+ fuel oil, asphalt or
feed to visbreaker
or coker.

The main products from a typical crude distillation unit are listed below in the

order of increasing boiling points (Gary, 1994).

12
 Fuel gas. TT.e fuel gas consists o f methane and ethane. The fuel gas is mainly

used as fuel for refinery furnaces. Sometimes it is also used as feedstocks for ammonia

production.

Wet gas. The wet gas consists o f propane and butane as well as methane and

ethane. The mixture o f propane and butane can be sold as LPG and butanes are raw

material for gasoline blending and some alkylation imit feed.

Naphtha. Most naphtha is used in gasoline production, either directly or as

catalytic reformate. Naphtha is also used as petrochemical feedstock, mainly for thermal

cracking to olefins or for reforming and extraction o f aromatics.

Gas oils. The light, atmospheric, and vacuum gas oils are used as feedstock for a

hydrocracker or catalytic cracker to produce gasoline, jet, and diesel fuels. The heavy

vacuum gas oils can also be used as feedstock for lubricating oil units.

Residue. The vacuum bottoms can be sent to a visbreaker, coker, or deasphalting

unit to produce heavy fuel oil or cracking and/or lube base stocks. For asphalt crude oils,

the residue can be used to produce road or roofing asphalt.

The relevant specifications o f these products are the following:

• ASTM D86 temperatures or D ll 86 temperatures (for vacuum gas oils)

• Flash point (for kerosene)

• Products 5-95 gap

The flash point o f a product is the temperature at which vapor given off will ignite

when an external flame is applied under specified test conditions. A flash point is defined

to minimize fire risk during normal storage and handling.

13
 The term gap refers to the difference between the 5% ASTM D86 distillation

temperature o f a heavier product and the 95% ASTM D86 distillation temperature of an

adjacent lighter product. When the distillation curves o f the two products overlap, a

negative gap appears.

1.4 State of the art in the design

One o f the most important early publications is “Petroleum Refinery Engineering”

by Nelson (1936). This book discusses the crude distillation in detail. It contains many

charts and experimental data. The 4* edition was published in 1958. In the design

procedure presented in this book, the amount of products is estimated from the crude

distillation curve, then the number o f trays between side withdraws are picked. Next, the

temperature and pressure in each tray are calculated using empirical charts. Heat balance

is then carried out and the reflux is determined. The design procedure ends with the

design o f the preheat train.

Another early publication is a paper by Packie (1941). In his method, 5-95 gaps

and 50% distillation temperature differences are used as separation criteria. Packie used

empirical charts to express the relation among the 5-95 gap, the reflux ratio, and the

number o f trays in the section under consideration. However, the empirical nature of

these charts results in inaccuracies in the amount of heat removal and tray temperatures.

Furthermore, as in the case o f Nelson (1936), Packie considered column design and heat

integration separately, that is, heat integration does not start until the column design is

finished.

14
 Much later, another book on crude distillation was published by Watkins (1979).

This procedure provided in this book is similar to the one proposed by Nelson (1958). It

uses empirical charts and iterations are required to compute heat and material balances.

However, departing from Nelson and Packie’s criterion regarding heat integration,

Watkins pointed out that “optimizing the crude preheat-tower cooling heat-exchange train

is the heart o f crude unit design, and each case must be studied on an individual basis in

order to arrive at the most economical processing scheme.” The statement reflects the

times the book was written (1979) right after the first energy crisis (1973). However, he

did not present a specific methodology to perform this integrated design.

Bagajewicz (1998a) studied the flexibility o f atmospheric crude distillation units.

He found that pump-around circuits are only effective to remove heat at higher

temperature. Main steam injection was found beneficial from the point o f view of energy

integration. The combined usage o f large steam injection with smaller overflash was

shown to have large energy savings. In another paper (1998b), he presented a technique

to calculate energy retrofit horizons in process plants. The methodology takes advantage

of two facts: (a) Pinch-type calculations can be performed using operator type

representations and (b) Processes like crude fractionation offer large flexibility in the

operating^^design parameters. The simulation package Pro/H (SimSci) was used to

perform these studies.

Recently, Liebmann et al. (1995, 1998) proposed an integrated design procedure.

The design procedure starts with a sequence of simple columns that are generated by

decomposing the crude main tower. The total number o f trays is assumed to be the same

as that o f Watkins’ design, and the number o f trays for each column is calculated with the

15
assumption that the R/Rmin values are approximately the same for all columns. There is

no thermal coupling between these initial columns. Next, reboilers and thermal coupling

are introduced in order to reduce utility consumption. To assess the proposed

modifications, a tool fi'om Pinch Technology (developed in the eighties), namely, the

grand composite curve, is used. After all the possible design modifications have been

explored, these columns are merged into a single complex column.

The major advantage o f the procedure presented by Liebmann et al. (1998) is that

it couples the column performance with heat recovery goals. However, the procedure is

not able to assess the trade o ff between steam injection and distribution of heat removal

in pump-around circuits. Finally, addressing the column as a whole, which is the

alternative we present in this paper, is more convenient and straightforward. It does not

rely on any special rules o f thumbs for reflux ratios, and it helps determine better the

relationships with column variables and heat integration.

Sharma et al. (1999) proposed a method for calculating the maximum pump-

around heat removal. First, a practical minimum reflux ratio for each column section is

determined using Packie’s empirical diagram. Then, the heat removal in the upper part o f

the column is calculated using a heat balance. The upper part may start fi'om an arbitrary

tray and end with the condenser. Following, the upper part is extended tray by tray and

heat siuplus is calculated for each tray. The resultant heat surplus data are used to

construct a column grand composite curve. Finally, the maximum heat removal for each

section is determined using the column grand composite curve. A major advantage o f this

method is that the maximum heat removal can be estimated quickly without the need o f

16
simulation. However, because Packie’s diagram is empirical and the effect o f the

stripping steam is not included, the heat removal calculated is not accurate.

To address all the aforementioned inaccuracies, it seems natural to think o f

simulator and heat integration procedures as means o f developing a new design procedure

that would capture rigorously the column behavior, the effect o f steam and refluxes and

the relationship with heat integration. Bagajewicz (1998b) suggested the use of rigorous

simulation for crude distillation design/retrofit.

The major features o f such procedure are:

• The plant should be able to process several crudes at optimal conditions.

• It should operate at maximum efficiency possible in all cases.

• The trade-off between energy and capital expenditure should be clear.

1.5 Topics in this dissertation

This dissertation discusses a rigorous procedure that uses simulators that are able

to rigorously represent the column operating conditions as a function of the design

parameters and complements it with state of the art heat exchanger network design

methods. In Chapter 2, a targeting procedure for conventional distillation will be

presented. This procedure uses a heat demand-supply diagram and commercial software

to determine the energy target for a plant processing several crude oils. This procedure

allows each crude to be processed at optimal conditions. Chapter 3 addresses the problem

o f designing a crude distillation with a pre-flash imit. A rigorous definition for “carrier

effect” is provided and the role o f light components in the crude distillation is analyzed in

detail. Then a rigorous procedure for energy targeting in the design o f crude distillation

with preflashing is provided. Chapter 4 is devoted to the study o f the stripping-type

17
distillation which was claimed by Liebmann and Dhole (1995) to be more energy-

efficient than the conventional design. The impact o f the different heating patterns in the

stripping-type design is therefore analyzed and compared with the conventional design.

Chapter 5 addresses the energy targeting for the design o f complete distillation plants,

that is, including both the atmospheric tower and the vacuum tower. Chapter 6 focuses on

the design of a multi-period heat exchanger network for a complete distillation plant. The

plant has a preflash drum to facilitate the processing o f light crudes. The last chapter

(Chapter 7) discusses the possible alternative designs and presents preliminary results.

1.6 References

1. American Petroleum Institute, 1970, Technical Data Book - Petroleum Refining, 2nd

Ed., Procedure 7B3.1, 7-29 - 7-286.

2. American Petroleum Institute, 1970, Technical Data Book - Petroleum Refining, 2nd

Ed., Procedure 7H2.1, 7-201 - 7-202.

3. Bagajewicz, M., On the design flexibility o f atmospheric crude fractionation units,

Chem. Eng. Comm., 166, 111-136, 1998a

4. Bagajewicz, M. J., Energy savings horizons for the retrofit o f chemical processes.

Application to crude fractionation units. Computers & Chemical Engineering. 1-9, 23

(1), 1998b

5. Gary, James H., Petroleum refining : technology and economics, 3rd ed. M. Dekker,

New York, 1994

6. Hans-Joachim Neumann, Petroleum Refining, Ferdinand Enke Publishers Stuttgart,

1984

18
7. Liebmann, K., Integrated crude oil distillation design. Ph. D. dissertation. University

of Manchester Institute o f Science & Technology, 1996.

8. Miller, W., Osborne, H., The science o f petroleum. Volume II, Oxford University

Press, London, 1938.

9. Nelson, W. L., Petroleum Refinery Engineering, 1* edition, McGraw-Hill, New

York, 1936.

10. Nelson, W. L., Petroleum Refinery Engineering, 4* edition. Chapter 4, McGraw-Hill,

New York, 1958.

11. Packie, J.W. Distillation Equipment in the Oil Refining Industry. AIChE

Transactions. 1941, 37, 51.

12. PRO/n Workbook, Hydrocarbon Distillation, Simulation Sciences Inc., 1996.

13. Rhodes, A. K., Worldwide refining report. Oil & Gas Journal, December, 37-86,

1993.

14. Watkins, R. N. Petroleum refiner)' distillation. G ulf Publishing Company: Houston,

1979.

19
 Chapter 2 Rigorous Targeting Procedure for the Design of

Conventional Atmospheric Crude Fractionation Units

2.1 Overview

The problem o f designing crude fractionation units is not only a distillation

design. It has the added complexity that these units should be able to process different

types o f crude, sometimes from heavy to light. Important heat exchange also takes place,

and the energy efficiency is related to the column design parameters. Part I o f this two-

part paper presents a rigorous targeting methodology to design this multipurpose plant,

which can be implemented using a commercial simulator. Part II deals with the heat

exchanger network design.

2.2 Introduction

Crude distillation is energy intensive. It consumes fuel in the equivalent of 2% o f

the crude processed. The conventional design (Figure 2.1), consisting o f a column with

side strippers and pump-around circuits, appeared 70 years ago (Miller, 1938), and is still

the design used in the refining industry. Watkins (1979) proposed a design procedure for

this system and discussed a few variants such as pump-back reflux and stripping using

reboilers. A few alternative designs can be found in the literature. For example, the

addition of pre-fractionation columns to this conventional design was proposed by

Brugma (1941) and is being used in several industrial sites. Another old design, the

carrier design, was proposed as early as the 1920s. This design makes use o f light

components to enhance the separation in the stripping section o f the column. Nelson

20
(1958) also mentioned some other alternative designs. All these old designs have been

abandoned for reasons that are not completely known or understood. One important fact

is that they were abandoned before the seventies, when energy consumption started to

play an important role in process economics. Because energy efficiency is now desired,

all these designs merit réévaluations. Nevertheless, the conventional design is widespread

and popular.

CONDENSER

w ater
• n a p h th a

PAl

PA2 kerosene

ste am
PA3 diesel

crude DESALTER
h e n (~
gas oil
sour w ater; steam
HEN FURNACE re sid u e

Figure 2.1 Conventional Crude Distillation

Crude is mixed with water and heated in a heat exchanger network before

entering a desalter where most o f the water containing the salt is removed. The desalted

crude enters another heat exchanger network and receives heat from hot streams. Both

heat exchanger networks make use o f the vapors o f the main column condenser, the

pump-around circuit streams, and the products that need to be cooled. The preheated

crude then enters the furnace, where it is heated to about 340-370 °C. The partially

21
vaporized crude is fed into the flash zone o f the atmospheric column, where the vapor

and liquid separate. The vapor includes all the components that comprise the products,

while the liquid is the residue with a small amount o f relatively light components in the

range o f gas oil. Tliese components are removed from the residue by steam stripping at

the bottom o f the column. In addition to the overhead condenser, there are several pump-

around circuits along the column, where liquid streams are withdrawn, cooled, and sent

back to upper trays. Products are withdrawn in liquid state from different trays and then

stripped by steam in side strippers to remove light components. Bagajewicz (1998)

offered a detailed discussion o f the effect o f the different variables on the energy

efficiency o f this conventional design.

Crude oil is a complex mixture. There exist about one thousand distinguishable

components with boiling temperatures varying from room temperature to over 550 °C.

Crude distillation yields mixtures called naphtha, kerosene, diesel, and gas oil. These

products are specified by ASTM D86 distillation temperatures.

Compared to common distillation o f discrete components, crude distillation has

the following specific features:

• Large processing quantity: The charge rate is the largest among all petroleum or

chemical processing units. The typical processing capability is around 15,000 per

day (100,000 bbl/day). In such a large-scale process, energy cost accounts for a larger

part o f manufacturing cost than in other processes.

• Large temperature variation throughout the column: The temperature difference

between the top tray and the flash zone is about 250°C, which means large heat

degradation throughout the column.

22
• Absence o f a reboiler: The main column functions as a rectifying section for products,

while side columns act as stripping sections.

• Low separation sharpness: Product quality is specified by ASTM boiling points

rather than component fractions as in the discrete component separation case. The

former is a more relaxed requirement.

• Components in a lighter product can be found in any o f heavier products: This is

because all components constituting the light product have to travel through trays

where heavier products are withdrawn.

The major objective in the design of crude distillation units is to find the most

energy efficient separation structure. Although some ideas exist for the design o f energy

integrated distillation schemes (Agrawal, 1996), they are not directly applicable to crude

fractionation for the following reasons:

• The number of components in the crude is too large to handle. Usually around 30 to

40 pseudo components are used, while available studies on sequencing seldom

addressed systems containing over 5 components.

• Previous separation sequencing studies assumed the products are pure, however

products in crude distillation are mixtures.

Packie (1941) pioneered the field o f crude fractionation design. In his method, 5-

95 gaps and 50% distillation temperature differences are used as separation criteria. The

term gap refers to the difference between the 5% ASTM D86 distillation temperature of a

heavier product and the 95% ASTM D86 distillation temperature o f an adjacent lighter

product. When the distillation curves o f the two products overlap, a negative gap appears.

Packie used empirical charts to express the relation among the 5-95 gap, the reflux ratio,

23
and the number o f trays in the section under consideration. However, the empirical nature

o f these charts results in inaccuracy and prevents optimal designs. Furthermore, in his

design procedure, Packie considered column design and heat integration separately. Heat

integration does not start until the column design is finished.

Watkins (1979) pointed out that “optimizing the crude preheat-tower cooling

heat-exchange train is the heart o f crude unit design, and each case must be studied on an

individual basis in order to arrive at the most economical processing scheme.” However,

Watkins did not present a specific methodology to perform this design.

Recently, Liebmann et al. (1995, 1998) proposed an integrated design procedure.

The design procedure starts with a sequence o f simple columns that are generated by

decomposing the crude main tower. The total number o f trays is assumed to be the same

as that o f Watkins’ design, and the number o f trays for each column is calculated with the

assumption that the R/Rmin values are approximately the same for all columns. There is

no thermal coupling between these initial columns. Next, reboilers and thermal coupling

are introduced in order to reduce utility consumption. The grand composite curve is used

to assess the proposed modifications. After all the possible design modifications have

been explored, these columns are merged into a single complex column.

The major advantage o f the procedure presented by Liebmann et al. (1998) is that

it couples the column performance with heat recovery goals. However, the procedure is

not able to assess the trade o ff between steam injection and distribution of heat removal

in pump-around circuits. Finally, addressing the column as a whole, which is the

alternative we present in this paper, is more convenient and straightforward. It does not

24
rely on any special rules o f thumbs for reflux ratios, and it helps determine better the

relationships with column variables and heat integration.

Sharma et al. (1999) proposed a method for calculating the maximum pump-

around heat removal. First, a practical minimum reflux ratio for each column section is

determined using Packie’s empirical diagram. Then, the heat removal in the upper part o f

the column is calculated using a heat balance. The upper part may start from an arbitrary

tray and end with the condenser. Following, the upper part is extended tray by tray and

heat surplus is calculated for each tray. The resultant heat surplus data are used to

construct a column grand composite curve. Finally, the maximum heat removal for each

section is determined using the column grand composite curve. A major advantage o f this

method is that the maximum heat removal can be estimated quickly without the need o f

simulation. However, as Packie’s diagram is empirical and the effect o f the stripping

steam is not included, the heat removal calculated is not accurate. In contrast, the

procedure presented in this paper is based on rigorous simulations and can capture the

relationships between the column variables and the heat integration opportunities.

In this paper we present a new procedure for crude distillation design. The major

features o f this procedure are:

• The design objective is to process several crudes at optimal conditions.

• The design calculations are rigorous.

Heat demand-supply diagrams are used as a tool guiding the design. The major

advantage o f the heat demand-supply diagram is that the contribution o f each process

stream or pump-around to the total utility consumption is shown explicitly. This feature

helps determine necessary changes leading to lower energy consumption.

25
• The interaction between steam stripping and pump-around duties is taken into

consideration.

• The starting point o f the column design is a complex column without pump-around

circuits.

While general procedures that would render globally optimal solutions are a

desirable goal, there is also interest in determining the optimal parameters for the subset

o f conventional units. This choice is made because practitioners are not always willing to

make radical departures from this design. In addition, the knowledge o f such optimal

designs provides a useful horizon for retrofit procedures. In addition, the heat exchanger

network design procedure presented in part II renders a structure suitable for the

processing o f different crudes at maximum energy efficiency. A short version o f this

procedure was presented by Bagajewicz et al. (1999).

The paper is organized as follows: The heat demand-supply diagram, an important

tool, is presented first and the roles o f the different column design variables in this

diagram are discussed. Next, limitations in the pump-around circuit heat load are

described. Finally, the design targeting procedure is presented. This rigorous procedure is

based on the use o f commercial simulators, departing from the use of charts, rules o f

thumb, and other approximations.

2.3 Heat demand-supply diagram

Heat demand-supply diagrams are an extension o f the concept o f temperature-

enthalpy diagrams (Hohmann, 1971; Huang and Elshout, 1976; Naka et. al., 1980;

Andrecovich and Westerberg, 1985; Terranova and Westerberg, 1989; Dhole and

26
LinnhofF, 1993). In the demand-supply diagram, a stream is represented by a curve. This

curve represents the product of mass flowrate and specific heat capacity (true or apparent

in the case o f phase changing streams) as a function of temperature. A schematic heat

demand-supply diagram for typical crude fractionation units, like the one o f Figure 2.1, is

shown in Figure 2.2.

In setting up the diagram, a heat demand line is first drawn and used as a

background. The crude is the only cold stream. In some cases, as proposed by Liebmann

et al. (1998), water at room temperature to produce steam is also considered a cold

stream. However, in many refineries, low-pressure steam is in surplus, and it can be

considered as a cheap or even free heat source. To locate the hot streams (heat supply),

the usual minimum temperature difference is used. Thus the temperatures o f hot streams

are shifted to the left by this minimum difference, which has been traditionally named

heat recovery minimum approximation temperature (HRAT). The area below the heat

demand line represents the total heat demand o f the unit without heat recovery.

In Figure 2.2, we see two different regions:

• Regions where heat supplies are larger than demands

• Regions where heat supplies are smaller than demands

When the supply exceeds the demand, one can move the surplus part o f the

supply to a lower temperature region where the supply is deficient. Figure 2.2 shows two

areas where the supply is in deficit (gray areas). The left gray area can be covered by the

heat surplus fi'om the condenser or from PA l. The right one can be covered using the

excess o f PA2. This illustration is omitted throughout the paper, assuming that this area

matching is implicit.

27
 A

.2

.0
o
o
PINCH
0.8 PA1
CRUDE
0.6 COND
Q.
o
*
0.4
PA2 PA 3
0.2
RES
0
100 200 300 400
PRODUCTS
TEMPERATURE, °C

Figure 2.2 Heat Demand-Supply Diagram o f a Crude Distillation Unit.

1. Naphtha and condensed water. 2. Sour water from the desalter. 3. Pump-around 1. COND: condenser.

PA: pump-around. RES: residue. Products (from top to bottom): kerosene, diesel and gas oil.

The location of the pinch point can be easily obtained from this diagram. It is the

lowest temperature at which the demand is larger than the supply after the shifting and

area matching has been performed. Finally, the heating utility is given by the unmatched

demand on the right, and the cooling utility is given by the extra supply on the left.

Three options to reduce energy consumption exist: decreasing demand, increasing

supply and improving the match between the supply and the demand.

28
• Decrease o f demand

a) A decrease in heat demand can be realized by moving the demand line down, that is,

decreasing the flowrate. One way o f doing this is to flash the crude at lower

temperatures and send the vapor to a tray above the flash zone. In practice,

vaporization before the furnace inlet is pressure suppressed to avoid two-phase flow.

This reduces energy saving opportimities. Such opportunity is analyzed in a separate

work (Ji and Bagajewicz, 2000).

b) Another way to decrease heat demand is to reduce the target temperature of the crude.

This can be achieved by lowering the pressure drop from the outlet o f the furnace to

the overhead reflux drum o f the column. In this sense, a vacuum operation is even

better, but it is excluded for other reasons (mainly cost). Another way of decreasing

the final temperature is using larger amount o f steam, but the introduction o f steam

has complicating effects on energy consumption, which will be discussed in another

paper.

• Increase o f the supply and/or the thermal quality

Withdrawing certain products in the vapor phase instead o f in a liquid phase has

the advantage that condensation heat is released at a higher temperature. This option is

not explored in this paper, mainly because it is a major departure from the conventional

scheme.

• Improvement o f match betw’een demand and supply

29
 Assume there is a large heat surplus in a moderate temperature range and a heat

deficit in a higher temperature range. One way to improve this mismatch is to move a

part o f the heat surplus to a higher temperature by increasing the duties o f the pump-

around circuits. The design procedure proposed in this paper relies on the idea of

distributing heat among the condenser and pump-around circuits. We focus on this

procedure next.

2.4 Pump-around circuits and heat recover)

The original purpose o f adding pump-around circuits was to reduce vapor and

liquid traffic at the top section o f the column (Watkins, 1979). Without pump-around

circuits, all condensation heat has to be removed from the condenser, which results in a

large vapor flowrate at the top trays. We explore now the limit o f heat that could be

removed fi-om a pump-around circuit.

Maximum Heat Duty’ o f Pump-Around Circuits

In Figure 2.3, envelope ///co n tain s A:pump-around circuits. In order to calculate

the maximum pump-around duty, we carry out a heat balance for this envelope.

«'S ■ + '-’b ■A& + Z.,_, ■ +z F „ ■ K +Z a

isl/l kç /II / 4 1V

= '%+Zie/II
+C ■A" + ■ K ,

In equation (1.1), V ^, are the steam flowrate at the flash zone and the

hydrocarbon vapor flowrate at the flash zone respectively, v f , y ° are the steam

flowrate at tray j and the hydrocarbon vapor flowrate at tray j respectively. In Figure 2.3,

30
we use Vjn = and Vj = V j +V f . It is assumed that water is insoluble in

liquid streams.

By applying material balance o f hydrocarbons to envelope I, one obtains:

(1 2 )
ie.1

Similarly, material balances for envelopes //a n d ///a re :

y" + iE
e // /
f.,
(1.3)
1'° = i . - . + Z f .
le//

Replacing V ^ , V j and V ° in equation (1.1), one obtains:

(/. ^- /«^) + 1 •( /'„ - A ° ) + 2 [ ■{h°- )

& € // / ( € /// /€ //
(1.4)
( A , T ( A ; : ' -A j+ /,_, (A °-A ,j

There are six terms on the right hand side o f equation (1.4). From left to right,

these terms represent condensation heat o f the overflash stream Lq, condensation heat of

the products leaving envelope III, apparent heat released by the hydrocarbon vapor V ° ,

and apparent heat released by the steam streams. The sixth term stands for tlie

vaporization heat o f internal reflux Lj.i. Apparently, when Lj.i goes to zero, the heat

removal from envelope III reaches its maximum. By including more pump-arounds in

envelope III and applying equation (1.4) accordingly, one can find the maximum heat

duty for each pump-around circuit.

It can be shown through an overall heat balance that the total amount o f heat to be

removed fi-om the column depends on the yields o f the products. In addition, shifting heat

31
from envelope I I to envelope III results in a decrease of Lj.i. Thus, the shifting can take

place as long as Lj.i remains positive.

^ Water
PAl Naphtha

SK
Kerosene
PA2
" III
SD
PA3 "► Diesel

SG
^ Gas oil
Hot Crude FZ
SR

Residue

V - . .- Figure 2.3 Heat Balance o f the Distillation Column

Effect o f Heat Shifting on Separation

It is well known that heat shifting reduces separation efficiency. The presence o f

the pump-around circuit decreases the number of effective ideal trays (Bagajewicz,

1998). The effect can be even more detrimental to separation if the flowrate of a pump-

around circuit is increased. As we shall see later, these effects can be compensated to a

32
certain extent by increasing the steam rate in the side strippers. Another solution is to

increase the number o f trays. However, with the total number o f trays kept constant, the

aforementioned relationship between heat recovery and steam consumption can be

incorporated into the design procedure.

2.5 Design procedure

We now summarize the technique for designing a multipurpose energy efficient

atmospheric column. First, the Watkins design method is used to obtain an initial scheme

without pump-around circuits. Then a heat demand-supply diagram is constructed, and

the direction o f heat shifting needed for maximum energy efficiency is determined. This

procedure is repeated for at least the lightest crude and the heaviest cmde that will be

processed. Thus, the design procedure is divided into two parts, the targeting procedure

and the multipurpose heat exchanger network design. This paper focuses on the targeting

procedure, which is presented next. After this, the goals o f the heat exchanger network

design procedure are outlined. The heat exchanger network design procedure is presented

in part II.

Step 1: Begin with the lightest crude to be processed. As the lightest crude has the

highest yields o f light distillates, the supply o f heat is the largest. Next, the major design

parameters (the number o f trays in each section, the pressure drop, and the amount o f

stripping steam) are chosen using the guidelines offered by Watkins with one exception:

No pump-around circuits are included at this point.

33
Step 2: The simulation is performed next. Usually the column is not difficult to converge,

as the liquid reflux ratio is large.

Step 3: The heat demand-supply diagram is constructed.

Step 4: The maximum amount o f heat is transferred to a pump-around circuit located in

the region between the top tray and the first product withdrawal tray. The location o f the

pump-around circuit withdrawal and the return temperature are conveniently chosen so

that the energy recovery is maximized. This is discussed further when presenting the

example.

Step 5: If the product gap becomes smaller than required, the stripping steam flowrate is

to be increased to fix the gap. As long as the steam added has a lower cost than the

energy saved, one can continue shifting loads. Otherwise, it is advisable to stop when a

trade-off has been reached.

Step 6: If there is heat surplus from the pump-around circuit just added, transfer the heat

to the next pump-around circuit between draws in the same way as in step 4. If not, stop.

At this stage, once this procedure is repeated for different crudes, one is left with

heat removal targets fi’om the condenser, the products and several pump-around circuit

streams. Typically, since the light crude is the one that needs a larger reflux, it exhibits a

larger amount o f pump-around circuit duties. After these targets are determined, it is

shown that there is still some flexibility to move heat fi’om one pump-around to another, a

feature that may be helpful in the final design o f the heat exchanger network, or for

retrofit. The above procedure is illustrated first. The results o f this targeting procedure are

34
used as motivating material to discuss the goals o f the multipurpose heat exchanger

network, which is presented in part II.

2.6 Illustration

The properties o f the light crude, intermediate crude and heavy crude are shown

in Tables 2.1, 2 and 3. Table 2.4 indicates the specifications of the products. The product

withdraw locations are detemiined according to W atkins’ guidelines and the results are

shown in Table 2.5.

Table 1 Feedstock Used for the Design

Crude Density Throughput
(kg/m^) (m^/hr)
Light Crude 845 (36.0 API) 795
Intermediate Crude 889 (27.7 API) 795
Heavy Crude 934 (20.0 API) 795

Table 2.2 TBP Data

Vol. % Light Crude Intermediate Heavy Crude
(°C) Crude (°C) (°(:)
5 45 94 133
10 82 131 237
30 186 265 344
50 281 380 482
70 382 506 640
90 552 670 N/A

35
 Tabic 2.3 Light-ends Composition o f Crude in volume per cent
Compound Light Crude Intermediate Crude Heavy Crude
Ethane 0.13 0.1 0
Propane 0.78 0.3 0.04
Isobutane 0.49 0.2 0.04
n-Butane 1.36 0.7 0.11
Isopentane 1.05 0 0.14
n-Pentane 1.30 0 0.16
Total 5.11 1.3 0.48

Table 2.4 Product Specifications and Withdrawal Tray

Product Specification Withdrawal Tray
Naphtha D86 (95% point) =182 °C 1
Kerosene D86 (95% point) =271 °C 9
Diesel D86 (95% point) =327 °C 16
Gas Oil D86 (95% point) =377-410 °C 25
Overflash rate 0.03
Kerosene -Naphtha (5-95) Gap > 16.7 °C
Diesel- Kerosene (5-95) Gap > 0 °C
Gas Oil- Diesel (5-95) Gap = -5.6 °C to -11 °C
Feed Tray 29
Total Trays 34

There are 34 trays in the main column and 4 trays in each stripper. The flowrates

of stripping steam streams are estimated and adjusted to 10 lb per barrel o f product, as

36
suggested by Watkins. The total energy consumption (£) is calculated using the

following expression:

E = U^0.1 * (2.5)

where U is the minimum heating utility obtained using straight pinch analysis, and

^ /// is the summation o f energy flow of all steam streams. Because low-pressure

steam is cheaper than fuel gas on the same amount o f heat content, a weight factor of 0.7

is used for the steam. The total energy consumption is used as an objective function.

Simulation results for the initial scheme with no pump-around circuits are shown

in Table 2.6. Note the product gaps are well above the specifications.

Table 2.5 Tray Requirements in Watkins Design

Separation Number o f Trays
Light Naphtha to Heavy Naphtha 6 to 8
Heavy Naphtha to Light Distillate 6 to 8
Light Distillate to Heavy Distillate 4 to 6
Heavy Distillate to Gas Oil 4 to 6
Flash Zone to First Draw Tray 3 to 4
Steam Stripping Sections 4

The heat demand-supply diagram corresponding to the solution in Table 2.6 is

shown in Figure 2.4. There is huge heat surplus in the condenser region, which results in

a large cooling utility. Meanwhile, a large heat deficit exists above 155 °C. As the total

heat supply is almost constant, the way toward energy savings is to change the heat

supply profile. That is, instead o f supplying all heat at a low temperature, some heat can

37
be supplied at a higher temperature where the heat demand is larger than the heat supply.

In other words, transfer some heat from the condenser to a pump-around circuit as

indicated by the arrow in Figure 2.4.

1.4

PINCH
O
COND
0.8 -

CRUDE
Q 0.6

^ 0.4 -

RES

100 200 300 400

PRO DUCTS
TEMPERATURE, °C

Figure 2.4 Heat Demand-Supply Diagram for Crude Distillation without Pump-Around

Circuits

One Pump-Around Circuit

If a pump-around is above all side-withdrawal product lines, the heat that can be

transferred from the condenser will be the maximum. Therefore, the first pump-around

38
has to be above the kerosene withdrawal tray. The question is how many trays one should

put between the condenser and the top pump-around region. We recommend the top

pump-around region be adjacent to the condenser. No tray is put in between. This is

based on the observation that the trays below a product withdraw line and above an

adjacent pump-around circuit receive little reflux and barely contribute to separation. The

pump-around stream is withdrawn from tray 4, cooled in the heat exchangers and

returned to tray 2. The return temperature is 104.4 °C, which is optimized after the duty is

determined.

The duty o f the top pump-around (PA l) is increased gradually and product gaps

are examined in each simulation. The kerosene-naphtha gap decreases with the increase

o f PAl duty, but remains well above that o f specification, while the other gaps are almost

unchanged. The heat shift continues without violating the gap specifications until the

reflux ratio is around 0.1. Further heat shift would result in liquid drying up on the top

tray. Thus, the limit o f the heat shifting has been reached. The duty of 62 MW represents

the total amount o f heat one could obtain from all pump-around circuits. The following

steps consist of distributing this amount of heat properly among several pump-around

circuits. The main operation variables of the scheme with one pump-around are shown in

Table 2.6.

The major conclusions are:

• The total energy consumption (E) decreases by 7 MW compared to the no pump-

around scheme.

• The kerosene-naphtha gap is reduced from 25 °C to 23 °C, remaining well above the

specification o f 16.7 °C.

39
Table 2.6 Comparative Results o f One Top Pump-Around and No Pump-Around
Product No Pump-around One Pump-around
Naphtha Flowrate 250 m^/hr 248 m^/hr
Kerosene Flowrate 144 m^/hr 146 m^/hr
Diesel Flowrate 70 m^/hr 70 m^/Tu"
Gas Oil Flowrate 121 m^/hr 121 m^/hr
Residue Flowrate 211 m^/lir 211 m^/hr
Kerosene Stripping Steam Ratio* 9.82 9.68
Diesel Stripping Steam Ratio 10.22 10.27
Gas Oil Stripping Steam Ratio 10.12 10.11
Residue Stripping Steam Ratio 10.19 10.19
Kerosene-Naphtha (5%-95%) 25.12°C 23.0°C
Gap
(5-95) Diesel-Kerosene Gap 5.14°C 5.3 r c
(5-95) Gas Oil- Diesel Gap 0.93 "C 0.91°C
Kerosene Withdrawal Tray 238.8°C 237. r c
Temperature
Diesel Withdrawal Tray 298.7°C 298.7°C
Temperature
Gas Oil Withdrawal Tray 338.7°C 338.7°C
Temperature
Residue Withdrawal Temperature 347.8°C 347.8°C
Condenser Duty 103.86 MW 41.70 MW
Condenser Temperature Range 155-43.3 °C 146.4-43.3 °C
Pump-around 1 Duty - 62.14 MW
Pump-around 1 Temperature - 179.6-104.4 °C
Range
Flash Zone Temperature 358.6 °C 358.6 °C
Energy Consumption (E) 103.78 MW 96.77 MW
*Steam amount in lb/hr over the amount of product in bbl/hr.

40
• The yield o f naphtha decreases and the yield o f kerosene increases. This is because

some light components of the vapor are absorbed by the cold pump-around stream

and carried to the kerosene withdrawal tray. Note that the total yield of the two

products remains constant.

• Little change takes place below the kerosene withdrawal tray.

We now turn our attention to the resulting heat demand-supply diagram (Figure

2.5). The shaded area is the energy savings achieved by adding PA l. The heat surplus in

the condenser region is greatly reduced, but it is still significant. However, it is

impossible to shift more heat from the condenser to PA l.

The return temperature o f PAl is not important in terms of energy consumption,

because the heat surplus is larger than the demand below the PA l withdrawal

temperature. To reduce the heat surplus in the region o f PA l, a second pump-around is

installed at a position as indicated in Figure 2.5.

The second pump-around (PA2) is positioned between tray 10 and tray 12, just

below the kerosene withdrawal tray. The return temperature is chosen to be

approximately equal to that of the withdrawal temperature of PAl. A lower temperature

would result in heat surplus in the region o f P A l, while a very high return temperature

would not alter the energy savings but result in a heavier liquid traffic in the PA2 region.

With the increase o f the PA2 duty, the gap between kerosene and naphtha decreases

quickly. Table 2.7 shows the change o f gaps as a function of the duty of pump-around

PA2.

41
 1.4

PINCH
O

0 .8 - PA1 CRUDE
& 06. -

0.4
COND
0 .2 -
RES

100 200 300 400

PRODUCTS
TEMPERATURE, °C

Figure 2.5 Heat Demand-Supply Diagram for Crude Distillation with a Top Pump-around
Two Pump-Aroimd Circuits

When the duty o f PA2 is larger than 33.7 MW, the kerosene-naphtha gap does not

satisfy the specification. To recover this gap, one could increase the stripping steam

flowrate or increase the number of trays in the naphtha-kerosene section. The former

option is used in this work, while the latter may not be sufficient or even practical on its

own. The kerosene and diesel stripping steam fiowrates are adjusted with a controller in

which the gap specifications are defined.

42
 Table 2.7 Effect o f Increasing PA2 Duty without Changing Steam Fiowrates

1 2
Duty o f PA2 29.31 MW 33.71 MW
Duty o f PA l 32.83 MW 28.43 MW
Duty o f Condenser 41.94 MW 42.03 MW
(5-95) Kerosene-Naphtha Gap 18.49 °C 16.60 °C
(5-95) Diesel-Kerosene Gap 1.63 °C 1.48 °C
(5-95) Gas Oil- Diesel Gap 1.22 °C 1.23 °C
Energy Consumption 70.59 MW 67.35 MW

With the help o f the stripping steam, it is possible to move more heat from PAl to

PA2. The trade off between increasing energy recovery and spending more steam is

evaluated using equation (2.5). Heat shifting continues until the liquid reflux at the

kerosene withdrawal tray is small and /or the kerosene-naphtha gap cannot be recovered

even with increased amounts o f stripping steam. This is a limit imposed by the separation

requirement. The limiting case is shown in Table 2.9 (first column) and should be

compared with the second column o f Table 2.6.

The major changes from one pump-around to two pump-around circuits are:

• The net energy consumption decreases sharply by 32 MW.

• The flowrate o f the kerosene stripping steam is nearly doubled. The large extra steam

is used to strip a significant amount o f light components in the kerosene withdrawal

stream. The top section o f the column becomes less hot because o f the increased

stripping steam. The kerosene withdrawal temperature drops by 33 °C.

• The yield o f diesel increases while the yield o f naphtha decreases.

43
 The heat demand-supply diagram (Figure 2.6) shows a good match, and the pinch

temperature increases to the value o f the PA2 withdrawal temperature. The heat surplus

in the region o f PA l is still high, but further shifting would cost too much steam to be

beneficial. Therefore, this remaining heat surplus is useless. Now the only heat surplus

transferable is located in the PA2 circuit, shown as the shaded area in Figure 2.6. To

make use o f this heat surplus, it is necessary to add a third pump-around circuit.

Three Pump-Around Scheme

The third pump-around (PA3) is located between tray 17 and tray 19. The return

temperature is 232 °C. Heat is shifted gradually from PA2 to PA3, with the gaps

maintained by adjusting steam fiowrates. The effect o f the duty of PA3 on energy

consumption is shown in Table 2.8. A summary o f all variables is given in Table 2.9.

Table 2.8 Effect of the Duty of PA3 on Energy Consumption

PA3 Duty (MW) Energy Consumption (MW )

6.45 61.96
8.79 61.64
13.19 61.67
23.45 63.76
26.67 64.56

44
 1.4

1.2

1 PINCH
o
g 0.8
CRUDE
5
cL 0.6
O COND
2 0.4

0.2

0
0 100 200 300 400

TEMPERATURE, °C
Figure 2.6 Heat Demand-Supply Diagram for Crude Distillation with Two Pump-Around
Circuits.

At the beginning, the energy consumption decreases with the increase o f the duty

o f PA3. However, when the PA3 duty exceeds 8.8 MW, the energy consumption stays

constant in a rather wide range (Table 2.8). This is because little heat surplus exists in the

region of PA2. Therefore, more heat shift makes no difference. Beyond this stable range,

more heat shift to PA3 results in an increase in energy consumption due to increased use

o f steam, which means that the cost of additional steam consumption outweighs the gain

in energy recovery. Clearly 8.8 MW is the right point to stop. This effect cannot be

captured with other design procedures.

45
 Table 2.9 Comparative Results for Two and Three Pump-Around Circuits

Product 2 Pump-arouiid 3 Pump-around

circuits circuits
Naphtha Flowrate 244 m^/hr 244 m^/hr
Kerosene Flowrate 145.6 m^/hr 145.5 m^/hr
Diesel Flowrate 73.6 m^/hr 72.5 m^/hr
Gas Oil Flowrate 121.6 m^/hr 123.85 m^/hr
Residue Flowrate 210.5 m^/hr 209.7 m^/hr
Kerosene Stripping Steam Ratio* 19.02 18.04
Diesel Stripping Steam Ratio 8.11 12.54
Gas Oil Stripping Steam Ratio 7.84 7.71
Residue Stripping Steam Ratio 10.20 10.24
(5-95) Kerosene-Naphtha Gap 16.7 °C 16.7 °C
(5-95) Diesel-Kerosene Gap 0 °C 0 °C
(5-95) Gas Oil- Diesel Gap -2.0 °C -2.9 °C
Kerosene Withdrawal Tray 202.2 °C 212.7 °C
Temperature
Diesel Withdrawal Tray Temperature 291.2 °C 289.9 °C
Gas Oil Withdrawal Tray Temperature 336.1 °C 338.9 °C
Residue Withdrawal Temperature 347.9 °C 348.2 °C
Condenser Duty 42.4 MW 43.3 MW
Condenser Temperature Range 143.6-43.3 °C 143.5-43.3 °C
Pump-around 1 Duty 22.3 MW 22.3 MW
Pump-around 1Temperature Range 169.2-104.4 °C 169.4-104.4 °C
Pump-around 2 Duty 42.5 MW 33.7 MW
Pump-around 2 Temperature Range 257.9-171.1 °C 255.3-171.1 °C
Pump-around 3 Duty - 8.8 MW
Pump-around 3Temperature Range - 310.6-232.2 °C
Flash Zone Temperature 358.7°C 359 °C
Energy Consumption 64.73 MW 61.64 MW

46
 1.4

PINCH

CRUDE
COND
0.4
PA2 PA3
0.2
RES

0 100 200 300 400

TEMPERATURE, °C

Figure 2.7 Heat Demand-Supply Diagram for Crude Distillation with Three
Pump-Around Circuits

Figure 2.7 is the heat demand-supply diagram. The heat surplus previously in the

region o f PA2 (Figure 2.6) has been moved to the PA3, which accounts for the decrease

in energy consumption.

It should be pointed out that the distribution of pump-aroimd duty has a

remarkable effect on the column temperature profile. Figure 2.8 shows stream

temperature changes as a function o f the duty o f the third pump-around. With the

47
increase o f these duties, the temperatures o f the products and the pump-around circuits

decrease.

350

Ü 300
o
UJ
a:
250
0:
LU
CL

UJ
h- 200

150

100
0 5 10 15 20 25 30
D U T Y . MW/ °C

Figure 2.8 Pump-Around Withdrawal Temperatures and Product Temperatures as a

Function of the Duty of the third Pump-Around (Light Crude)

Finally, the effect of the pump-around return temperatures is briefly explored. As

the pinch temperature is located in the region o f PA2, its return temperature affects the

energy consumption. Figure 2.9 shows this effect.

48
 1.2

1
PAl
0.8
CRUDE

0.6
O
o COND

0.4
Q. RAI
U
♦
0.2 PA2
RES
0

0 100 200 300 400

TEMPERATURE, °C

Figure 2.9 Effect o f the Return Temperature o f PA2 on Energy Consumption

In the figure, PA3 is not shown for simplicity. The dotted line is the new PA2

region with a higher return temperature and the shaded areas are the changes incurred.

When the PA2 duty is constant, a higher return temperature results in a lower withdrawal

temperature. The total effect depends on the trade off between these two effects, or the

area difference between the shaded triangle and the shaded rectangle. The trade off is

shown in Table 2.10. The optimal return temperature is found to be 177.8 °C.

49
 Table 2.10 Effect o f the Return Temperature o f PA2 on Energy Consumption

PA2 Return Withdrawal Diesel Stripping Energy

Temperature Temperature Steam Consumption
(°C) (°C) (MW) (MW)
171.1 255.3 2.15 61.64
177.8 254.4 2.17 60.91
193.3 251.9 2.25 61.96

At this point we have reached the best scheme for the light crude. Next we

perform the same analysis for the heavy crude.

Heavy Crude

The total energy consumption and the pump-around duty distribution are shown

in Table 2.11. The heat demand-supply diagram and the operation variables for a scheme

with three pump-around circuits are shown in Figure 2.10 and Table 2.12. The following

results are observed:

Table 2.11 Effect o f Pump-Around Duties on Energy Consumption

(Heavy Crude, AT= 5.6 °C)
PAl Duty PA2 Duty PA3 Duty Energy
(MW) (MW) (MW) Consumption
(MW)
0 0 0 24.27
6.10 0 0 23.88
2.32 4.34 0 23.88
2.32 2.20 2.14 23.79

50
• The energy consumption changes very little when shifting heat from the condenser to

the pump-around circuits, especially when heat is shifted from PAl to PA2 or PA3.

• This is because that there is no heat surplus in the condenser region (Figure 2.10).

However, because the light crude and the medium crude require the PA2 and PA3

heat exchangers, shifting heat from PAl to PA2 and PA3 in heavy crude design may

be necessary.

0.8

0.6 CRUDE
PA3
Ü
o
0.4

RES
O
Q. COND
♦ 0.2
PA1
SW PA2

0
100 200 300 400
TEMPERATURE, °C

Figure 2.10 Heat Demand-Supply Diagram for Heavy Crude Distillation

51
 Table 2.12 Results for Heavy Crude
Product Heavy Crude
Naphtha Flowrate 55.37 m^/hr
Kerosene Flowrate 48.64 m^/hr
Diesel Flowrate 69.36 m^/hr
Gas Oil Flowrate 29.37 m^/hr
Residue Flowrate 592.51 m^/hr
Kerosene Stripping Steam Ratio* 1.63
Diesel Stripping Steam Ratio 2.98
Gas Oil Stripping Steam Ratio 37.9
Residue Stripping Steam Ratio 2.68
(5-95) Kerosene-Naphtha Gap 26.07 °C
(5-95) Diesel-Kerosene Gap 0.86 °C
(5-95) Gas Oil- Diesel Gap -5.84°C
Kerosene Withdrawal Tray Temperature 259.7°C
Diesel Withdrawal Tray Temperature 317.4 °C
Gas Oil Withdrawal Tray Temperature 344.4 °C
Residue Withdrawal Temperature 366.7 °C
Condenser Duty 14.8 MW
Condenser Temperature Range 123.3-18.5 °C
PAl Duty 20.8 MW
PAl Temperature Range 175.7-104.4 °C
Flash Zone Temperature 353.2 °C
Energy Consumption 81.49 MW

When heat is shifted to PA2 and PA3, more steam is needed for the diesel stripper to

regain the kerosene-diesel gap. The diesel stripping steam fiowrates for the designs with

one pump-around, two pump-around and three pump-around circuits are 32.5, 48.5 and

52
113.4 kg-moie/hr respectively. Although the steam consumption increases, the total

energy consumption is barely affected because the heat from the extra steam is utilized to

cover the heat deficit in the condenser region.

• The separation o f kerosene and diesel in the column is much easier than that o f the

light crude. Before stripping, the gap between kerosene and naphtha is 17.2 °C,

satisfying the separation requirement.

2.7 Effect of minimum temperature approach

The effect o f HRAT on the optimal pump-around duty distribution is shown in

Tables 13, 14 and 15.

Table 2.13 Effect o f HRAT on Energy Consumption (Light Crude)

HRAT PA l Duty PA2 Duty PA3 Duty Energy Consumption

(°C) (MW) (MW) (MW) (MW)
5.6 22.3 34 8.8 61
22.2 22.3 29 8.8 69.8
44.4 22.3 23 13.2 81.2

Table 2.14 Effect o f HRAT on Energy Consumption (Heavy Crude)

HRAT PA l Duty PA2 Duty PA3 Duty Energy

(°C) (MW) (MW) (MW) Consumption
(MW)
5.6 7.9 7.5 7.3 81.2
22.2 22.6 0 0 86.4
44.4 22.6 0 0 93.1

53
 Note that for the light crude, PA3 duty increases with the increase o f HRAT. This

can be explained using the heat demand-supply diagram (Figure 2.7). When the HRAT is

5.6°C, there is almost no heat surplus in the region o f PA2. However, when HRAT is

increased, the crude demand curve is moved to the right and heat surplus appears again.

Thus, the heat surplus needs to be reduced to achieve the maximum energy savings. The

heavy crude behaves differently. As there is no heat surplus in the region of the

condenser and P A l, shifting heat fi'om PA l to PA2 or PA3 does not reduce the net heat

demand while more stripping steam is needed to keep the product gaps. At low HRAT

(e.g., 5.6 °C), most o f the heat coming from the condenser can be used because o f the

heat deficit in the condenser region. However, when HRAT is raised, the overlapping

between the crude curve and the condenser curve reduces, and part o f the heat from the

condenser is at a temperature that is too low to be usable. In such a case, the heat from

the increased steam cannot be used. Therefore, heat shifting to the lower pump-around

circuits is not beneficial.

Table 2.15 Effect of HRAT on Energy Consumption (Medium Crude)

HRAT PAl Duty PA2 Duty PA3 Duty Energy Consumption

(°C) (MW) (MW) (MW) (MW)
5.6 15.2 26.4 0 63.1
22.2 15.2 26.4 0 70.4
44.4 15.2 26.4 0 79.9

These calculations were also performed for the intermediate crude (Table 2.15).

In this case, the heat distribution does not change with HRAT This is because there is

always a heat surplus in the region of PA l and a heat deficit in the region o f PA2. The

54
heat surplus in the region o f PA l prompts maximum heat shift to PA2, while the heat

deficit in PA2 excludes the need for shifting heat to PA3. Thus, the optimal solution is to

maximize the duty o f PA2.

2.8 Flexibility

The basis o f this design procedure is the transferring o f heat duty from the

condenser to the lower pump-around circuits. In doing so, limits to this transfer are

encountered. In addition, heat surplus can be transferred back. For example, the heat

surplus observed in the pump-around PA l for the light crude (Figure 2.7) could in

principle be transferred back to the condenser without affecting the utility consumption.

Such flexibility is further discussed in part II.

2.9 Reboiler

Reboilers are rarely used in the conventional crude distillation because their

installation is expensive and sometimes they are less efficient than steam stripping.

However, Liebmann et al. (1998) suggested that the use o f reboilers can lead to energy

savings. This conclusion is made by analyzing a scenario where water to produce steam

is considered as another cold stream. We now revisit this analysis.

If one assumes that steam is another heat sink, then the location o f the pinch

changes. Figure 2.11 shows the demand-supply diagram of the light crude. The heavy

solid curve represents the heat demand for both the crude and the steam. The heat

demand for the crude is shown as a broken curve for comparison. In this figure, a heat

surplus can be observed in the PA2 region.

55
 1.4 -1
1.2 - PINCH
I

o 1 -
o
0.8 -
i DEMAND
CL 0.6 - COND
o 1
"K 258°C
0.4 H
182 °C
0.2
0 -
300
200 400
0 K E ^O
TEMPERATURE, C
igure 2.11 Heat Demand-Supply Diagram for Light Crude Including Steam as Cold
Stream

To utilize the heat surplus, there are three options:

• Shift the heat to the third pump-around.

• Produce low-pressure steam.

• Install a reboiler for stripping kerosene.

Each option has its advantage. Shifting the heat surplus to the third pump-around

can reduce the duty o f the furnace. Heat provided by the furnace is relatively expensive,

because the furnace is usually less efficient than a boiler. However, heat shifting from

PA2 to PA3 deteriorates the separation and consequently, extra steam is needed for

56
stripping gas oil. Thus, the choice between the first two options is a matter o f a cost-

beneflt analysis, and it can not be determined merely on energy savings considerations.

The choice between producing steam versus installing a reboiler depends on the

amount o f light components to be stripped. The use o f steam is more efficient for

stripping a low amount o f light components, while a reboiler is better for removing a

relatively large amoimt o f components (Nelson, 1958, Liebmann et ai, 1998). It is also

possible that a combination o f these options offers an optimal solution. To find the best

design, all options should be carefully evaluated.

When steam is considered free or is priced as it is proposed in this article, without

allowing its production to participate in the design procedure, then the installation of

reboilers should be analyzed using the demand-supply diagram offered in Figure 2.6 or 7.

In this case, the shifting o f heat to PA3 is energy savings related and the use of reboilers

needs to be made by using some other external heat source because process heat is no

longer available in this case. Thus, their installation can only influence the steam

consumption but it is replaced by a similar duty. The choice is therefore not driven by

energy savings, but by a cost-benefit analysis that depends on the cost o f the reboilers to

install and their efficiency relative to direct steam.

2.10 Applicability to other crudes

The light crude and the heavy crude represent two extremes in the raw material

spectrum. Any other crude could be thought o f as a mixture o f the two crudes. Based on

the above information, we may design an optimal HEN that allows several crudes being

processed at optimal conditions. The conjecture is that building a network addressing the

57
extremes allows also processing a crude o f intermediate density at maximum energy

efficiency. This is explored in more detail in part II.

2.11 Conclusion

In this paper, a rigorous targeting design procedure has been proposed for the

design o f conventional crude distillation units. This procedure is an improvement over

the existing procedures for several reasons. First, this procedure aims at finding the best

scheme for a multipurpose crude distillation unit that processes a variety o f crudes.

Second, the procedure is more straightforward. Heat demand-supply diagrams, instead o f

grand composite curves are used as a guide directing the search for optimal schemes. An

advantage o f heat demand-supply diagrams is that the role of each stream, heater or

cooler in the total energy consumption is clearly shown, so the search of the best scheme

is straightforward. Third, the approach is rigorous. The trade o ff between different

operating parameters is considered and the decision is based on quantitative calculations

instead o f simple assumptions. The second part paper concentrates on the design o f heat

exchange network.

2.12 Nomenclature

CR = (hot) crude oil

E = energy consumption defined by equation (5), MW

F = mass flowrate, kg/'s

F^. = mass flowrate o f product i, kg/s

F^.- mass flowrate o f steam i, kg/s

58
/ / / = enthalpy o f stripping steam i, MW

h'^ = enthalpy o f water (steam) at the flash zone, kJ/kg

= enthalpy o f hydrocarbon vapor at the flash zone, kJ/lcg

= enthalpy o f liquid falling from trayy-1, kJ/kg

= enthalpy o f liquid falling into the flash zone, kJ/kg

hp. = enthalpy o f product /, kJ/kg

= enthalpy o f steam z, kJ/kg

hy = enthalpy o f water (steam) rising from tray j , kJ/kg

h ° = enthalpy o f hydrocarbon vapor rising from tray j, kJ/kg

L^ = overflash rate, kg/s

Q* = duty o f pump-around circuit k

R = reflux ratio

Rmin = minimum reflux ratio

RES = Residue

S = steam

SD = diesel stripping steam

SG = gas oil stripping steam

SK = kerosene stripping steam

SR = residue stripping steam

V'^ = water (steam) flowrate at tray j

Vp2 = vapor flowrate at flash zone

59
 = water (steam) flowrate at flash zone

V ^ = hydrocarbon vapor flowrate at flash zone

Vj = vapor flowrate at tray j

V f = hydrocarbon vapor flowrate at tray y

U = minimum heating utility excluding steam, MW

2.13 References

1. Agrawal, R., Synthesis o f Distillation Column Configurations fo r a Multicomponent

Separation. Ind. Eng. Chem. Res. 35, 1059-1071, (1996).

2. Andrecovich, M., and Westerberg, A., A Simple Synthesis Method Based on Utility

Bounding fo r Heat-integrated Distillation Sequences, AIChE Journal, 31(3), 363-375,

(1985).

3. Bagajewicz, M., On the Design Flexibility o f Crude Atmospheric Plants. Chemical

Engineering Communications, 166, 111-136, (1998).

4. Bagajewicz, M., S. Ji and J. Soto. A Step-By-Step Targeting Procedure fo r The Design

o f Conventional Crude Distillation. Enpromer 99, Florianopolis, Brazil, September.

Submitted to Latin American Applied Research (1999).

5. Brugma, A. J., The Brugma Process. Refiner and Natural Gasoline Manufacturer,

20(9), 86,(1941).

6. Dhole, V. R. and Linnhoff, B., Distillation Column targets. Computers and Chemical

engineering, 17(5/6), 549-560, (1993)

60
7. Glinos, FC., and Malone, M. F., Optimality Regions fo r Complex Column Alternatives

in Distillation Systems. Chem. Eng. Res. Des., 66, 229-240, (1988).

8. Hohmann, E. C , Optimum Networks fo r Heat Exchangers. Ph.D. Thesis, University

o f Southern California, (1971).

9. Huang, P., and Elshout, R., Optimizing the Heat Recovery o f Crude Units. Chemical

Engineering Progress, 72(7), 68-74, (1976).

10. Ji S. and M. Bagajewicz. Rigorous Targeting Procedure For The Design O f Crude

Fractionation Units With Pre-Flashing. Escape 10. Florence, Italy (2000).

11. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &

Chemical Engineering, 19, Supplement ,S119,(1995)

12. Liebmann, K.; Dhole, V. R. and Jobson, M., Integration Design O f A Conventional

Crude Oil Distillation Tower Using Pinch Analysis. Institution o f Chemical

Engineers, 76(3), part A, 335-347, (1998).

13. Linnhoff, B., Dim ford, H., and Smith R., Heat Integration o f Distillation Columns

into Overall Process. Chemical Engineering Science, 38(8), 1175-1188, (1983).

14. Miller, W. and Osborne, H. G. History and Development o f Some Important Phases

o f Petroleum Refining in the United States. The Science of Petroleum (Oxford

University Press, London), Volume 2, 1465-1477, (1938).

15. Nelson, W. L., Petroleum Refinery Engineering, 4 ^ ed.. New York, (1958).

16. Packie, J.W., Distillation Equipment in the Oil Refining Industry. AIChE

Transactions 37, 51-58. (1941).

61
17. Naka, Y., Terashita, M., Hayashiguchi S., and Takamatsu T., An Intermediate

Heating and Cooling M ethod fo r a Distillation Column. Journal o f Chemical

Engineering o f Japan, 13(2), 123-129,(1980).

18. Sharma, R., Jindal, A., Mandawala, D., and Jana, S. K., Design/Retrofit targets o f

Pump-around Refluxes fo r Better Energy Integration o f a Crude Distillation Column,

Ind. Eng. Chem. Res. 38, 2411-2417, (1999)

19. Teiranova, B., and Westerberg, A., Temperature-Heat Diagrams fo r Complex

Columns. 1. Intercooled/Interheated Distillation Columns. Ind. Eng. Chem. Res. 28,

1374-1379, (1989)

20. Tedder, D.W., The Heuristic Synthesis and Topology o f Optimal Distillation

Networks. Ph.D. Thesis, University o f Wisconsin, Madison, (1975).

21. Watkins, R. N., Petroleum Refinery Distillation. G ulf Publishing Company, (1979).

62
 Chapter 3 Rigorous Targeting Procedure for the Design of

Crude Fractionation with Pre-flashing

3.1 Overview

This paper presents a systematic procedure to obtain design targets for heat

integration in conventional crude fractionation units that use pre-fractionation columns or

pre-flash drums. It is shown that under the same high product yield conditions, pre-

fractionation or pre-flashing are not advantageous from the energy point of view. This is

in great part due to the loss o f the carrier effect that light components have in separating

heavy gas-oil fractions in the flash zone. However, if one accepts the yield of

atmospheric gas oil to be smaller, then these pre-fractionation/pre-flash options consume

less energy.

3.2 Introduction

In previous work (Bagajewicz and Ji, 2001), a systematic procedure for the design

of conventional crude fractionation units was presented. This procedure is based on a

step-by-step combination o f rigorous simulation and heat integration. The procedure

starts with a column without pump around circuits and as heat is transferred from the

condenser to pump around circuits with higher temperature, a trade off between steam

usage and furnace savings is established. This transfer o f heat is possible due to the well-

known operating and design flexibility that crude fractionation installations exhibit,

knowledge that was formalized in detail by Bagajewicz (1998). The procedure presented

by Bagajewicz and Ji (2001) makes use o f rigorous simulations and heat supply-demand

63
diagrams similar to those introduced by Andrecovich and Westerberg (1985) and

Terranova and Westerberg (1989). Based on these targets, a heat exchanger network

design procedure to handle crudes o f different density at maximum efficiency was

developed (Bagajewicz and Ji, 2001; Bagajewicz and Soto, 2001). These design

procedures did not use any pre-flashing or pre-fractionation schemes.

In this paper, a rigorous procedure to obtain targets o f energy consumption and

column reflux arrangements for pre-ffactionation/pre-flashing units is presented. Figures

3.1 and 3.2 depict both schemes.

3 __ condenser

PAl
heavy naphtha
light naphtha
steam
Crude oil
desalter steam
diesel

M
sour water

furnace ► residue
prefractionation
tower mam tower

Figure 3.1 Basic Pre-fractionation Design (Heat exchanger network is omitted).

In the pre-flash scheme (Figure 3.2) the intended effect is to avoid unnecessary

heating o f light components in the furnace, short-circuiting them to be injected at an

appropriate tray in the column for further fractionation. In the pre-fractionation scheme

64
 (Figure 3.1), light products obtained in the pre-fractionation column are not sent to the

main column.

condenser

naphtha
PAl

steam
PA2
Crude Oil
vapor
c kerosene

desalter steam
PA3 ^
diesel

steam
sour water gas oil
preflash drum steam
furnace residue

Figure 3.2 Basic Pre-flash Design (Heat exchanger network is omitted).

There are several unresolved questions regarding these designs. Among others:

1. In the case of pre-flashing, what is the optimum temperature o f the flash drum?

2. In the case o f pre-fractionation, what is the optimiun temperature of the pre­

fractionation column feed?

3. In the case o f pre-flashing, what is the feed tray that one needs to use to feed the

vapor from the drum into the column?

4. What are the loads o f the pump-aroimd circuits and what are the steam flowrates for

stripping that produce the most energy efficient combination?

In addition, both schemes avoid introducing lights in the feed tray (flash zone)

with the aforementioned expectation that one would avoid heating these lights

unnecessarily. However, the so-called carrier effect o f lights is known to increase the

65
separation o f atmospheric gas-oil from the residue (Golden, 1997). Since steam can act as

a carrier too, the question is whether it can actually substitute for the light hydrocarbons.

Finally, if yield is to be sacrificed, what is the extent o f yield reduction compared to

energy savings? In other words, would it be worth reducing the yield to achieve energy

efficiency?

All these questions have to be answered taking into account the fact that a specific

quality o f products (given by TBP or ASTM D86 points) needs to be guaranteed equally

for all options that are compared.

3.3 Effect of pre-flashing on products

In this section the step-by-step procedure proposed by Bagajewicz and Ji (2001) is

applied to both pre-fractionation and pre-flash schemes. In the case o f pre-fractionation,

the method can be applied directly. In the case o f pre-flashing, the introduction o f the

vapor feed from the flash drum changes the heat load distribution in the column affecting

the heat duty distribution o f the pump around circuits. Thus, to better study the effect o f

the drum temperature and the feed tray, a column without pump around circuits is first

used. Pump-around circuits are systematically added later.

The main column has 34 trays and side products are withdrawn at trays 9, 16 and

25 respectively. The outlet temperature of the furnace is set at 360 °C and the condenser

temperature is fixed at 32 °C. Two crude oils, one light and one heavy are used in the

study. Details o f the assay data and product 95% distillation temperatures were given in

by Bagajewicz and Ji (2000). The over flash rate chosen is 3%, the flowrates o f side

withdrawals and the condenser duty are adjusted to achieve the specifications o f the over

66
flash rate and product 95% distillation temperatures and the flowrates o f stripping steams

are maintained constant. Finally the minimum approach temperature for energy targeting

(HRAT) is 22.2 °C (40 T ) , as suggested by Bagajewicz and Soto (2001)

Consider first a pre-flash drum operated at 163 °C and the feed tray to be tray 15.

Tray 15 is chosen because the vapor concentration best matches the composition o f the

vapor from the drum. The effect o f changing these two values will be studied later. The

simulation results corresponding to a light crude with no pump-around circuits are shown

in Table 3.1 and Figure 3.3. As shown by Bagajewicz (1998), pump-around circuits have

an effect on energy recovery, but a very slight effect on product yield. Thus, as we shall

see later, the energy consumption shown in Table 3.1 should further decrease as pump-

around circuits are added.

Table 3.1 Comparison between Conventional and Pre-flash Design (Light crude)

Conventional Preflashing
Naphtha, M^/hr 249.6 249.3
Kerosene, M^/hr 144.1 143.9
Diesel, M^/hr 70.5 69.7
Gas oil, M^/hr 118.6 101.9
Residue, M^/hr 212.5 230.6
Product gaps, C
Naphtha-Kerosene 25.1 24.5
Kerosene-Diesel 5 3.2
Diesel-GO 0.6 -0.9
Heating utility, MW 103.6 97.3
Energy consumption, MW 113 106.8

67
 10000

V A P O R FEED
^ 8000

i UQUID FEED
O 6000

uT
^ 4000
K
O 2000
^ 1
---------------------------------- ---------- ,-----------

10 20 30 40
TRAY

LIQUID F L O W R A T E : NO PREFLASH ▲ PREFLASH

V A P O R F L O W R A T E :----------------- N O P R E F L A S H ♦ PR E F L A S H

Figure 3.3 Comparison o f Vapor and Liquid Distribution

The major differences are:

1. The vapor and liquid traffic decrease as a result of decreased heat input in the feed.

That is, a lower reflux is needed.

2. The yield o f gas oil decreases and the yield o f residue increases as a result o f less

stripping ends existing at the flash zone o f the column, and

3. The gaps between the different products decrease.

As we shall see later, the difference in yield for the gas-oil and the residues cannot

be reconciled. Essentially, the presence o f lights in the flash zone for the conventional

case, which provide the so-called carrier effect, cannot be substituted by other stripping

means in the case o f the pre-flash or pre-fractionation designs. The comparison with a

68
pre-fractionation design is omitted as the same effect is expected for the reasons just

outlined.

The same comparison was performed for the case o f a heavy crude (Table 3.2). In

this case no pump around circuits were used either.

Table 3.2 Comparison between Conventional and Pre-flashing Design (heavy crude)

Conventional Pre flashing

Naphtha, M^/hr 54.5 54.2
Kerosene, M^/hr 47.9 48.4
Diesel, M^/hr 71.1 70.8
Gas oil, M^/hr 28.8 26.4
Residue, M^/hr 593.0 595.4
Product gaps, C
Naphtha-Kerosene 19.8 18.6
Kerosene-Diesel 1.3 1.0
Diesel-GO -5.6 -5.1
Heating utility, MW 101.3 100.5
Energy consumption, MW 106.5 105.8

Specifically, the yield of gas oil decreases. As the heavy crude contains less light

distillates, vaporization in the heat exchanger network train is not severe and can be

suppressed under moderate pressure. On the other hand, the heavy crude flashing at a low

temperature (e.g., 163 °C) does not produce much vapor. From the viewpoint o f energy

savings, a small amount o f vapor bypassing the furnace cannot significantly reduce the

duty o f the furnace. We can see from Table 3.2 that the energy consumption for the pre­

flash design is only slightly lower. Therefore, the pre-flash design is not justified for

heavy crude. The same can be said for the pre-fractionation design.

69
 We now proceed to determine the optimal pump-around circuit loads. The heat is

shifted from the condenser to the pump around circuits step-by-step, as proposed by

Bagajewicz and Ji (2001). Briefly, the method consists o f transferring as much heat as

possible from the condenser to the pump around circuits until the overall minimum

utility consumption reaches a minimum. The procedure is summarized next.

Step 1: Start with a column configuration suggested by Watkins with one exception: No

pump-around circuits.

Step 2: Perform a simulation.

Step 3: Construct the supply-demand diagram.

Step 4: Transfer the optimum amount o f heat to a pump-around circuit located in the

region between the top tray and the first draw.

Step 5: If needed, increase the steam in the first side-stripper until the desired gap is

restored.

Step 6: If there is heat surplus from the pump-around circuit just added, transfer the heat

to the next region between draws in the same way as in step 4, if not, stop.

The fact that pre-flashing reduces gas oil yield reveals that the light components

in the crude help the vaporization o f heavy components. However, the yield of gas oil can

also be increased by using more stripping steam. In other words, both light components in

the crude oil and stripping steam have a stripping effect. We now discuss both cases:

• The stripping agent is steam.

• The stripping agent is a hydrocarbon, which exists in the mixture to be stripped.

70
 Table 3.3 Comparison between Conventional and Pre-flash Designs with pump around
circuits (light crude)
Conventional Preflashing
Naphtha, M^/hr 246.6 245.4
Kerosene, M^/hr 143.1 142.6
Diesel, M^/hr 73.9 74.4
Gas oil, M^/hr 119.4 102.4
Residue, M^/hr 212.5 230.6
Product gaps, C
Naphtha-Kerosene 20.5 18.8
Kerosene-Diesel -1.9 -6.6
Diesel-GO -0.3 -1.7
Heating utility, MW 84.2 78.4
Energy consumption, MW 93.7 87.9

In the rest o f the article, we will first explore the carrier effect and its intricacies

in the case of crude fractionation, and then present a comparison between the pre-flash

design and the conventional design for processing the light crude oil on the basis that

both designs produce the same amount of residue.

3.4 Carrier effects

For the purpose o f our analysis, the stripping effect o f a component is defined as

the differential change in the amount of residual liquid over the differential change of the

amount of this component in the feed, provided that the system temperature, pressure and

the quantities o f other components are constant. Thus, when the addition of a component

to the feed decreases the liquid rate, the assumption is that it happens at the expense o f

71
vaporizing more o f the heavy or intermediate components. Translated into the crude

fractionation field, the presence of light ends, decreases the residue yield and therefore,

increases the gas oil yield at its expense. We now analyze both stripping agents

separately.

Case 1: Steam stripping

Stichlmair and Fair (1998) present a few charts obtained numerically, showing

that the liquid yield in a flash is lowered when light components are added to a mixture.

We attempt to explain this theoretically. We start using the well-known flash equations:

X: =

(3.1)

where x, and z, are the molar fraction o f component i in the liquid and in the feed, is the

saturated pressure o f component i, P is the total pressure o f the system and L and F are

the liquid flowrate and the feed respectively. The set o f equations (I) is referred to as the

no-steam equation. The vapor-liquid phase behavior is assumed to follow Raoult law,

which is appropriate for hydrocarbons.

Let z ' , y ’ be the dry compositions and F ‘ , F ‘ the dry flowrates. The temperature

is assumed to be high enough so that any water in the liquid is neglected. A simple

balance tells that:

72
 (3.2)

where Pst is the partial pressure o f steam, and P ' is the system pressure. For atmospheric

distillation, P* is somewhere between 0,24 and 0.37 MPa. Since P ' -y, = P^^ x ., we

obtain:

ü P-
(3.3)
X* P -P ..

For convenience, we define AT*=-^. AT* is called the modified vapor-liquid

equilibrium constant o f component i. Finally, combining a dry component balance

F ;.z'= L-x]+ V '-y::

z.
X i =

+ (1 -
F F
2 ^ x ’, =1 (3.4)

P-
K. =
' P'-P..

We now compare (1) and (4). For a given hydrocarbon mixture flashing at a fixed

temperature, we have z ’ = z,. and F ' = F . Suppose the flash without steam and the flash

with steam take place at the same temperature, then the saturated pressure o f component i

does not change. When P= P ’ - Pst, we have K] = K- and x* = x ,. This means that

injecting steam is equivalent to reducing the system pressure. Note that equation (3) still

holds if steam is replaced by a gas that is insoluble in the hydrocarbon mixture. The use

o f steam, however, has some operational limits that will be explored later.

73
Case 2: Hydrocarbon stripping

We now consider the stripping effect o f light components that distribute in both

phases. Thus, using the following flash equations

I, = ---- t A (3.5)

i = Z ^ (3.6)
j

where /, is the flowrate of component i in the liquid phase.

Differentiate (5) and consider that dF ! = 1, to get:

' d l '' (3.7)

Because the liquid-phase partition ratio l/fi is less than I, the summation in

equation (7) is less than 1. Accordingly, the denominator is positive. Therefore, whether

the derivative is negative or not depends on the nominator. The nominator consists of two

terms. The first term represents the liquid-phase partition ratio of component / in the

liquid phase, the second term is the average partition ratio o f the mixture. If the partition

ratio o f component i is smaller than the average partition ratio, the derivative is negative,

and the component has stripping effect. Thus, equation (7) can be used to determine if at

a given pressure and temperature, a component is a stripping agent, that is, increasing its

composition in the feed leads to a decrease o f the liquid flowrate in the flash.

The stripping effect of each component in a mixture of aliphatic hydrocarbons is

studied next. The composition o f a mixture and the K.-values of each component are

shown in Table 3.4. The mixture is allowed to flash at 0.1 MPa and 393 °C. Figure 3.4

74
shows the residue/ feed ratio as a function o f K-values o f the added components. The

molar ratios o f the pure component added to the feed are 1/100, 5/100, 10/100, 20/100

and 30/100, respectively.

Table 3.4 Mixture Composition and K-values

Molar Composition K-Value

Water 0 267
Ethane 0 138
Propane 0 115
n-Butane 0 85.6
n-Hexane 0 45.2
n-Octane 0 27.4
n-Decane 0 17.8
n-Dodecane 0.04 12.2
n-Tetradecane 0.053 8.68
n-Hexadecane 0.064 5.84
n-Octadecane 0.077 3.79
n-C20 0.089 2.56
n-C22 0.098 1.75
n-C24 0.108 1.14
n-C26 0.117 0.79
n-C28 0.121 0.54
n-C30 0.107 0.38
n-C32 0.081 0.26
n-C36 0.045 0.13

Originally, the residue/feed ratio is 0.43. The residue/feed ratio decreases with an

increasing K-value o f the added component. Note that all curves intersect at K=1.35, at

75
which (— pr equals zero. When the K-value of the added component is greater than
a/, -

1.35, the residue/feed ratio is lower than 0.43. This means these components act as

stripping components. With the increase o f the K-value, the residue/feed ratio decreases.

When the K-value is greater than 30, the residue/feed ratio is almost constant. In this

sense, water, ethane, propane and butane have the same stripping effect on a molar basis.

1.4
1/100 5/100
1.2
o ............. 10/100 X 20/100
I— 1
o — * - — 30/100 — - — - 0/100
0.8 X \
X «
Transition at (1.35, 0.43)
LU 0.6
ID
Q
C/D 0.4 ♦ ♦ ♦ ♦ ♦ ♦♦♦ ♦
LU
CXL \ X X J ................................................... ..

0.2 ^ X x x x x X X X X

0
0.1 1.0 10.0 100.0 1000.0
K

Figure 3.4 Stripping effect as a function o f K-values

(Cio through C 36 mixture, 0.1 MPa, 393 °C)

At the same time, it can seen that for a given component with K value greater than

1.35, the residue/feed ratio decreases with an increasing amount o f the added component.

When K-value is less than 1.35, the residue/feed ratio is higher than that without stripping

76
component. This means that the component added is not a stripping agent. Instead, it

prevents vaporization.

The phenomena can be explained as follows. An added component helps the

vaporization by increasing its partial pressure, but it also hurts vaporization o f other

components by reducing their liquid molar compositions. Furthermore, the part of the

component distributed in the liquid phase will contribute to a larger residue/feed ratio.

When the K value o f a component is large, it generates a large partial pressure, and most

o f it stays in the vapor phase, so the net effect is stripping. On the other hand, when the K

value is small or in other words, the component is heavy, a large portion of the

component stays in the liquid phase, resulting in a larger residue/feed ratio.

To determine if this effect manifests quantitatively in the same approximate way

for crudes, the light crude was mixed with stripping agents and flashed at 360 °C. The

heavy crude was mixed with stripping agents and flashed at 343 °C. The results are

shown in Figure 3.5 and 6, which depict similar trends to that seen in Figure 3.4. The

residue/feed ratio in light crude stripping, however, does not go down as much because o f

the existence of a large amount o f light components in the mixture. The effect is shown

parametric as percentages, which are the amount of stripping agent divided by the amount

o f the crude on molar basis.

In practice, the crude experiences flashing at the flash zone of the main tower, and

the resultant liquid is further stripped with steam at the bottom part of the main tower. To

show how light components affect the flowrate of the residue leaving the main tower, we

split the vapor from the pre-flash drum (163 °C) into two parts as shown in Figure 3.7.

77
 One part goes to the main tower and the other mixes with the remaining crude

again and resultant mixture enters the furnace. Now concentrate on the furnace and

examine the effect o f the light components entering into it. The outlet temperature of the

furnace is fixed at 360 °C. By changing the remixing ratio, one can control the amount of

light components entering the furnace. The result is shown in Figure 3.8.

g 0.8
^20%
I 0.6 — 0%

g 0.4
o
CO
g 0.2

0 1 10 100 1000

K-VALUE

Figure 3.5 Stripping effect for light crude

The amount o f residue increases constantly with an increasing vapor separation

ratio. The vapor remixing ratio of one corresponds to the conventional design, and the

ratio o f zero corresponds to the common pre-flash design. It is seen that the yield of the

residue decreases constantly with the increasing vapor remixing ratio. The trend is

consistent with what we saw in Figure 3.4 through 6.

78
 Following, we first discuss the effect o f steam temperature and the location of its

injection. We tlien investigate the energy requirement associated with the use of

stripping agents, and lastly we study the effect of process pressure.

1.2

1. 1 10%

^30%
O 1
UJ 0%
liJ
u_ 0 .9
ÙJ
3
Q
(Ô 0.8
cc
0.7

0.6
0.1 1 10 100 1000

Figure 3.6 Stripping effect for heavy crude.

Vapor to the main tower

Desalted crude

To flash zone

preflash drum
furnace

Figure 3.7 A schematic diagram for pre-flash vapor splitting

79
 2 4 0 .0

preflashing

230.0
CO
s
UJ
3
2 conventional
S 220.0

210.0
0 0.2 0.4 0.6 0.8 1
V A PO R SEPARATION RATIO

Figure 3.8 Residue yield as a function of vapor separation ratio in pre-flash design

3.5 Steam Stripping: Effect of the temperature

Typically, there is a temperature drop of about 17 °C between the flash zone and

the bottom tray of the atmospheric column (Watkins, 1979). The temperature drop is

related to the vaporization of relatively light components being directly stripped by the

steam. The question is whether a higher steam temperature reduces the residue yield. As

the atmospheric column does not allow large flowrate o f steam, which would result in the

formation o f free water on top trays, we use the flowchart o f Figure 3.9 to perform this

study.

80
 Oil Vapor

Desalted 360 C
Crude

HEN Furnace

Steam

Residue

Figure 3.9 Flowsheet for Studying the Effect o f Light Components

The stripping power o f steam as a function o f flowrate at different temperatures is

depicted in Figure 3.10.

To show the limit o f the stripping power, the value o f the steam/feed ratio is

allowed to reach high values. Some o f these values are unrealistic, but they help illustrate

the point. When the steam/feed ratio is small, the temperature o f the steam does not affect

the yield of residue because the heat steam extracted from the oil is very small. When the

steam/feed ratio is raised to about 40, the amount of heat taken by the steam becomes

significant, and the 177 °C steam is worse than the 260 °C steam due to a lower

temperature profile at the stripping section of the column. On the 177 °C steam curve, a

minimum point is obser\ ed at a steam/feed ratio o f 75. This means that above this ratio,

the cooling effect o f the steam dominates and the yield o f residue goes up with an

increasing amount o f stripping steam.

81
 360

320 Steam 260 C

u. steam 177 C

a I 280
S 2

II 240

200 •

160
0 100 200 300 400 500 600

Steam/Feed ratio, LB/BBL

(a)

240

steam 260 C
200 steam 177C

160 •

I
120

80
0 100 200 300 400 500 600

Steam /Feed ratio, KG/M^

(b)

Figure 3.10 Residue yield and temperature as functions o f the flowrate o f stripping steam

(a) Residue flowrate (b) residue temperature

82
3.6 Steam Injection

In this section, we discuss two cases: injecting steam into the cnide before the

furnace and after the furnace. Figure 3.11 is used for the simulation. The flash zone

pressure is assumed to be the same as that o f the outlet o f the furnace. The results are

shown in Table 3.5.

Table 3.5 Effect o f steam injection location

Steam/feed, Residue for SBF*, Residue for Flash zone temp, for SAF*, C
KG/M^ M^/hr SAF*, M^/hr
0.0 207.3 207.3 360.0
3.3 202.7 204.1 359.4
5.5 199.8 202.0 359.1
11.0 193.0 197.2 358.3
22.1 181.6 188.6 357.0
*SBF= steam injected at the inlet o f the furnace. SAF = steam injected at the outlet of the furnace.

V apor

Desalted 360 C
Crude

HEN Furnace Sicam after the

furnace
Steam before
the furnace
Steam

Residue

Figure 3.11 Flowsheet for Studying the Effect o f Steam Injection

83
 In both cases, the yield o f residue decreases with the increased steam/feed ratio.

For the same steam/feed ratio, however, injecting before the furnace produces less

residue. This is because in this scheme, the steam-crude mixture leaving the furnace has

achieved a vapor-liquid equilibrium and no temperature change takes place between the

outlet of the furnace and the flash zone. In the other case, the hot crude, which is vapor-

liquid mixture at 360 °C, meets with the injected steam (also 360 °C). Because o f the

stripping effect, more hydrocarbons vaporize. As the vaporization process is adiabatic,

the temperature o f the system goes down. The last column in Table 3.5 shows the

temperatures in the flash zone. Therefore, injecting steam before the furnace is more

advantageous.

3.7 Energy requirements for stripping agents

We showed that substances with large K-values have the same stripping effect on

molar basis. However, the energy requirement for heating a gas to an elevated

temperature varies. From the viewpoint o f energy efficiency, gases requiring the lowest

energy are the best. Figure 3.12 shows heating curves for water, Ni, H: and light

hydrocarbons. Water has the largest heating requirement due to the heat o f vaporization.

Nitrogen and hydrogen have the smallest energy demands, and light hydrocarbons lie in

between, with the energy demand increasing with their molecular weight. Therefore,

nitrogen and hydrogen are the best stripping agents. The replacement o f steam by Nz or

Hz has the advantage that corrosion related to water formation in the distillation system is

reduced or even eliminated. The drawback is that they need to be separated from the

overhead product and recycled back. The flowsheet containing Nz or Hz stripping is

84
therefore slightly more complex, and an economic analysis is needed to determine

whether they are economically more beneficial than the conventional steam stripping.

60

50

40
0
30
1 20

10 . -▲

0
0 100 200 300 400 500
TEMPERATURE, C
CH4.g — - — - H2 N2
■C02 ^ ^ "^ ^ w a te r ■ethane, g
H P ro p a n e , g — — N-butane.g
-

Figure 3.12 Enthalpy increase as a function o f the final temperatures

(base: 25 °C, 1 atm)

3.8 Effect of pressure

The atmospheric tower operates at a pressure in the range o f 0.2-0.3 MPa. Figure

3.13 shows the heating curves for n-butane and N412, which is a pseudo component with

a molecular weight of 167.1, at 0.1 MPa and 0.26 MPa, respectively. The figure shows

that pressure only shifts the temperature at which vaporization takes place. It is also seen

that on a weight basis, the heat requirements for components with large difference in

molecular weight are very close.

85
 2000

♦ C 4 ,0 .IM P a
X N 4 1 2 ,0 .IM P a *
1500
- - - C4.0.26 MPa
s
N412,0.26 MPa
X
K 1000

M
s
500 X

JKL
100 200 300 400 500

TEMPERATURE, C

Figure 3.13 Effect of pressure on heating curves

From the above study, we reach the following conclusions:

Steam stripping enhances vaporization by lowering the partial pressure o f the

hydrocarbon vapor.

The stripping effect o f a hydrocarbon not only depends on its property, but also

depends on the property o f the mixture to be stripped.

The temperature o f the bottom stripping steam does not affect the yield of the residue

when the steam/feed ratio is less than 40.

Injecting steam before the furnace produces less residue than injecting after the

furnace.

86
• Nz and Hz and light hydrocarbons have the same stripping ability as water but require

less heat to be heated up. From the standpoint o f energy savings, they are better than

water.

• Pressure does not affect the heating curve except postponing the vaporization.

Finally, we can naturally envision two ways to increase the vaporization ratio at a

fixed temperature. One is to put more light components into the crude, which can be

achieved by adding stripping gases or recycling light distillates such as naphtha or

kerosene. The other way is to remove heavy components from the crude. We leave the

investigation of these points for the future.

3.9 Comparison for lower AGO yield

We now compare the pre-flash and pre-fractionation options with the

conventional option at lower AGO yields. Table 3.6 summarizes the result. In this

comparison, the stripping steams were adjusted so that both designs have the same

amount o f residue and the same product gaps.

The heat demand-supply diagrams corresponding to the optimal solution are

shown in Figure 3.14. Not all the heat possible to transfer from the second pump-around

(PA2) to the third (PA3) was transferred because the trade o ff between this load and the

increased steam consumption created a trade-off.

87
 Table 3.6 Comparison between Conventional, Pre-flash and Pre-fractionation Designs
(light crude)
Conventional Preflashing Pre fractionation
Naphtha, M^/hr 246.1 245.5 222.5
Kerosene, M^/hr 144.5 144.8 148.8
Diesel, M^/hr 70.7 71.6 71.5
Gas oil, M^/hr 103.6 102.9 105.7
Residue, M^/hr 230.5 230.6 227.5
Product gaps, C
Naphtha-Kerosene 19.9 19.1 18.2
Kerosene-Diesel 0.4 0.0 -0.4
Diesel-GO -2.5 -1.9 -1.7
Heating utility, MW 82.2 76.0 80.4
Energy consumption. 92.5 87.7 90.4
MW
* Pre-flash temperature: 163 °C. Vapor feed at tray 15. AT=22.2 °C.

It is shown that energy consumption o f the pre-flash design is 4.8 MW lower than

that of conventional design. Thus, the energy consumption of the pre-fractionation design

lies between the two designs. To understand the results, two aspects have to be taken into

consideration: one is the temperature level o f the heat to be recovered and another is the

amount of components bypassing the furnace. As the pre-fractionation column operates at

a low-temperature where the heat supply is in surplus, distributing heat between the pre­

fractionation condenser and the main column condenser does not affect the heating utility

requirement. Therefore, the only aspect affecting the energy consumption would be the

amount of components bypassing the furnace. When the temperature of the feed entering

the pre-fractionation column is the same as that entering the pre-flash drum, the amount

88
of components bypassing the furnace in the former would be lower because a part of the

vapor entering the column is condensed as the reflux and then mixes with the crude at the

bottom o f the pre-fractionation column. In terms o f energy consumption, such a pre­

fractionation design would be equivalent to a pre-flash design operating at a lower

flashing temperature.

Note that product distributions vary considerably from one design to another,

although the product gaps are exactly the same. For example, the pre-flash design gives

more diesel and less gas oil compared to the conventional design. This might be

preferable in some cases.

0.8 PA1

o
0.6 COND CRUDE

CL 0.4
o PA3
*
0.2 PA2
RES

0 100 200 300 400

TEMPERATURE, 0
Figure 3.14 Final Heat Supply-demand Diagram for a Pre-flash Design (AT= 22.2 °C)

C: condenser. PAl, PA2 and PA3: pump-around circuits. RES: residue.

89
3.10 Effect of pre-flash tem perature

The necessity of a pre-flash drum depends on the property o f the crude, the

temperature o f the crude entering the furnace and the operating pressure. The bubble

temperatures o f the lighter crude are higher than that of the heavier crude (Figure 3.15).

The higher the operating pressure, the higher is the bubble temperature. If the

temperature o f a crude oil exceeds its bubble temperature considerably before entering

the furnace, a pre-flash drum would be necessary.

300

250

200
UJ
Q.
S A P I= 3 6 .0
UJ
»- 150
A P I= 3 2 .8

100
0 0.5 1 1.5 2
PRESSURE, MPa

Figure 3.15 Crude Oil Bubble Temperature as a Function o f Pressure

The effect of the pre-flash drum temperature on the low AGO yield arrangement

is shown in Figure 3.16. With the increase o f the pre-flash temperature, the heating

utility decreases continuously because o f more light components bypassing the furnace.

90
On the other hand, the steam consumption increases rapidly as more steams are needed

for both keeping the gas oil yield and fixing the product gaps. Although the total energy

consumption reaches a minimum at 177 °C, the differences in the curve are relatively

small. However, if steam is in surplus in the refinery, heating utility only counts and

lower temperatures should be used.

110

h 50
Total energy consum ption
100
- 40
S

O 90 H eating utility - 30
a:
ill
Z
UJ 20
80 Steam consum ption
- 10

70
160 180 200 220
TEMPERATURE, C

Figure 3.16 Effect o f the pre-flashing temperature on total energy consumption

3.11 Feed location of pre-flash drum vapors

The vapor produced in the pre-flash drum at 163 °C has properties very similar to

naphtha. It contains 53% light ends (C 2 -C 5 ) Its 98% boiling point is 169 °C, lower than

91
the 95% temperature o f naphtha (182 °C). We compared the effect of locating this feed

in different trays in the column. During the comparison steam flowrates have been kept

constant, allowing the gaps to change. The results are given in Table 3.7.

Table 3.7 Effect of Vapor Feeding Locations

Tray 8 Tray 15 Tray 18 Tray 27 Tray 29

Naphtha, M^/hr 249.5 249.3 249.3 249.3 249.5

Kerosene, M^/hr 143.8 143.9 143.5 143.4 143.9

Diesel, M^/hr 69.6 69.7 70.3 68.6 68.7

Gas oil, M^/hr 101.9 101.9 101.8 103.5 102.3

Residue, M^/hr 230.6 230.6 230.6 230.6 231.0

Product gaps (°C)

Naphtha-kerosene 24.9 24.5 24.5 24.5 24.9

Kerosene-Diesel 3.8 3.2 2.1 2.5 4.0

Diesel-gas oil -0.9 -0.9 -0.8 -4.7 -0.9

We have the following observations:

1. The naphtha yield and the gap between naphtha and kerosene are constant. This is

because the vapor feed is basically naphtha components and the naphtha-kerosene

separation section (tray 1 to tray 9) is almost not affected in the above feeding

locations.

2. When the vapor feed location is lower than that o f the gas oil withdrawal (tray 25),

the yield o f diesel is significantly lower and the yield o f gas oil is higher. This is

92
 because that the temperature of vapor feed is much lower than that o f the vapor rising

from the flash zone and a portion o f the hot vapor is condensed with the effect of

some diesel components going to the gas oil draw stream. In Table 3.7, we can see

when the vapor feed tray is 27, not only the yield o f diesel decreases but also the

diesel-gas oil gap moves in the negative direction. Therefore, vapor feed between

flash zone and gas oil withdrawal tray is not appropriate.

3. When the vapor feed is sent to a tray above the flash zone, the residue yield is

constant. This is because that the residue yield depends on four factors: the

composition and temperature o f the crude at the outlet of the furnace, the flowrate of

stripping steam entering from the bottom o f the main column, and the over flash rate.

In the above situations, all the four factors are the same.

380
-T R A Y 8

18
360 - tra y

Ü A TRAY 27

UJ X TRAY 29
a: 340
z>
:
LU
CL 320
2
LU
h-
300

280
15 17 19 21 23 25 27 29 31 33
TR AY

Figure 3.17 Temperature Profiles for Various Vapor Feeding Locations

93
 The temperature profiles are shown in Figure 3.17. The tray temperatures are

highest when the vapor is fed at tray 8, the highest location. When the vapor is fed at tray

18, the tray temperatures above the vapor-feeding tray are lower, which is due to the

cooling effect o f the low temperature-feeding vapor. The tray temperatures below the

vapor feed tray are exactly the same as that when the vapor is fed at tray 8. When the

vapor feed location continues to move down, the same tendency appears as expected.

5000
i
—I TRAY8
0
s 4000 + TRAY 27

§ ^ ^ A A TRAY29
3000

1
“h

2000
▲
1000
+ ■A’ -4 r -4 r T
0
24 26 28 30 32 34
TRAY

Figure 3.18 Vapor Profiles as a Function of Vapor Feeding Tray

Finally, when the vapor location reaches its lowest location — the flash zone (tray

29), the temperatures o f tray 29 through tray 34 are much lower than all the others with

other vapor feed location. At the flash zone, the hot crude flashes and splits into rising hot

vapor and descending hot unstripped residue. The hot vapor meets with the cold pre-

94
flashed vapor immediately and condensation occurs. The condensed liquid, basically gas

oil and diesel components, mixes with the unstripped residue on the tray 29. However,

the condensed liquid does not stay with the unstripped residue and is stripped out when

the mixture of the unstripped residue and the condensate meets the steam coming from

tray 30.

In Figure 3.18, we can see that the vapor flowrate o f tray 29 is about twice as

large as that when the pre-flashed vapor is fed at other locations. The overall effect is

equivalent to that o f conventional design with lower outlet temperature of furnace. It can

be predicted that heating o f the pre-flashed vapor to higher temperature would reduce the

temperature drop at tray 29 and help increase the output o f distillates.

From the above comparison, we can conclude that product yields and gaps are not

sensitive to vapor feed location. Specifically, the residue yield keeps constant. As a

higher temperature profile is favorable in heat recovery through pump-around circuits,

higher vapor feed is preferred. However, as pointed by Golden (1997), a higher vapor

feed would have the risk o f contaminating the products withdrawn below the feed tray

when liquid oil entrains.

3.12 Conclusion

A rigorous targeting for the design of a crude distillation unit with pre-flash

drums and pre-fractionation is presented. The major conclusion o f this study is that pre-

flashing and pre-fractionation designs can only be advantageous from the point of view

o f energy consumption if one accepts the yield of gas oil to be reduced. Other findings of

95
the study show that in practice, the equivalent o f the carrier effect o f lights cannot always

be obtained using steam.

In addition, the introduction o f a flash drum into an already complicated heat

exchanger network needs to be considered carefully. For example, it might be

advantageous to pay the penalty o f higher energy consumption just to see the number o f

units in the preheating train reduced. Finally, heat efficiency and network simplicity is

desired for a variety o f crudes simultaneously. This, together with measures to improve

the performance o f these units will be analyzed in future work.

3.13 Nomenclature

fi = flowrate o f component i in the feed

F = feed flowrate, kgmol/hr

F* = dry flowrate of the feed, kgmol/hr

A", = vapor-liquid equilibrium constant for component i

AT*= modified vapor-liquid equilibrium constant for component i

L = liquid flowrate, kgmol/hr

/, = flowrate o f component i in the liquid phase

Pst = partial pressure o f steam in the vapor phase

P = total pressure o f the system without steam

P*= total pressure o f the system with steam

P ' = saturated pressure o f component i at the system temperature

V = vapor flowrate, kgmol/hr

V*= vapor dry flowrate, kgmol/hr

96
X, = molar fraction o f component i in the liquid phase

x ' = dry molar composition o f component i in the liquid phase

>•, = molar fraction o f component i in the vapor phase

_v* = dry molar composition o f component i in the vapor phase

z, = molar fraction o f component / in the feed

z* = dry molar composition o f component i in the feed

3.14 References

1. Andrecovich, M., and Westerberg, A., A Simple Synthesis Method Based on Utility

Bounding for Heat-integrated Distillation Sequences. AIChE Journal, 31(3), 363-375,

(1985).

2. Bagajewicz, M., On the Design Flexibility of Crude Atmospheric Plants. Chemical

Engineering Communications, 166, 111-136, (1998).

3. Bagajewicz M. and S. Ji. Rigorous Procedure for the Design of Conventional

Atmospheric Crude Fractionation Units Part 1: Targeting. Industrial and Engineering

Chemistry Research. Vol. 40, No 2, pp. 617-626 (2001).

4. Bagajewicz M. and J. Soto. Rigorous Procedure for the Design of Conventional

Atmospheric Crude Fractionation Units Part II: Heat Exchanger Networks. Industrial

and Engineering Chemistry Research. Vol. 40, No 2, pp. 627-634 (2001).

5. Golden, S., Prevent Pre-flash Drum Foaming, Hydrocarbon Processing, May 1997,

ppl41-153.

6. Stichlmair, J.G, and Fair J. R, Distillation Principles and Practices, Wiley-VCH, New

York, 1998.p76.

97
7. Terranova, B., and Westerberg, A., Temperature-Heat Diagrams for Complex

Columns. 1. Intercooled/Interheated Distillation Columns. Ind. Eng. Chem. Res. 28,

1374-1379,(1989).

98
 Chapter 4 On the Energy Efficiency of Stripping-Type Crude

Distillation

4.1 Overview

In this chapter, the energy efficiency o f crude fractionation using stripping-type

columns is analyzed. Rigorous simulations prove that in the case o f processing a light

crude of reference, previous reports indicating that this option is advantageous over the

conventional design cannot be generalized.

4.2 Introduction

In conventional crude atmospheric distillation, the crude is fed at the bottom part

of the column, and the main part of the column functions as a rectifying section. Watkins

(1979) suggested the possibility of using direct and indirect rectifying sequences. The

conventional design that he chose to describe in detail is shown in Figure 4.1. Its

connection to a indirect sequence of distillation columns is shown in Figure 4.2. In recent

years, the alternative indirect sequence (Figure 4.3) was proposed by Liebmann (1995).

In previous work (Bagajewicz and Ji, 2001, Bagajewicz and Soto, 2001), a

systematic procedure for the design o f conventional crude fractionation units was

presented. This procedure is based on a step-by-step combination o f rigorous simulation

and heat integration. The procedure starts with a column without pump around circuits.

When heat is transferred from the condenser to pump around circuits, a trade off between

the steam usage and the fuel gas savings is established. This transfer of heat is possible

due to the operating and design flexibility that crude fractionation installations exhibit.

99
Such design flexibility was studied in detail by Bagajewicz (1998). The procedure makes

use o f rigorous simulations and heat supply-demand diagrams similar to those introduced

by Andrecovich and Westerberg (1985), and Terranova and Westerberg (1989). Based on

these targets, a universal heat exchanger network which allows several crudes to be

processed with the minimal energy consumption can be found (Bagajewicz and Soto,

2001 ).

^CONDENSER
~\J~ ^wjiter
> naphtha

PAl

steam
PA2 kerosene

steam
PA3 diesel

water — T
crude— DESALTER steam
HEN g asoil
sour water!
— I steam
'
V
HEN FURNACE >■ residue

Figure 4.1 Conventional Crude Distillation

100
 Condenser

-► Naphtha

Kerosene

-► Diesel

Crude _► Gas oil

Residue

Figure 4.2 Indirect sequence (Watkins, 1979)

In the stripping-type design, the crude is heated to a relative low temperature

(about 150°C) and fed at the top o f the column. Because the crude temperature is low, the

vapor ratio of the feed is small. The crude goes down the column and is heated

consecutively in three heaters (the upper heater, middle heater and lower heater). Side

products are withdrawn from the vapor phases, and rectified in the side rectifiers.

Liebmann and Dohle (1995) reported that this design could feature 5% less utility

cost than the optimized conventional design. However, the comparison was not on the

same allowable temperature basis. The maximum allowable temperature is limited by the

thermal stability property o f the crude being processed and is found by lab testing. For

101
the Venezuela crude oil used in Liebmann’s paper, the allowable temperature is 343 °C

(Watkins, 1979). However, in their stripping-type design, the crude was heated to 370 °C

in both the middle heater and the lower heater. It can be expected that at this temperature,

severe thermal cracking takes place. Such operation is not allowed in practice. Because o f

the above limitation of their study, a re-evaluation o f the stripping-type design is

necessary. This chapter performs the evaluation using the method proposed by

Bagajewicz and Ji (2001), taking into account the temperature limit of thermal cracking.

Preheater ^ Water
Crude oil
j^. -» Naphtha
”0 Kerosene
UH
-► D iesel

MH
"► Gas oil

LH

Steam
Residue

Figure 4.3 Stripping Type Crude Distillation Column

102
4.3 C rude vaporization patterns and beat dem and

A major difference between the conventional design and the stripping-type design

is the heating pattern o f the crude oil. Because in the stripping-type design crude oil is

heated step-by-step and the vapor is separated immediately after it is generated, the

amount of crude to be heated in the stripping section is smaller than in the conventional

design. On the contrary, the amount of crude being heated in the conventional design is

constant, that is, no vapor is separated until the heating is completed at the outlet o f the

furnace.

To better understand the difference between the two heating patterns, one should

analyze the distillation curves. It is well known that distillation curves can be used to

approximate the distillation behavior in terms of product distribution and vaporization

ratio. The two relevant distillation curves, ASTM D86 and EFV, are useful tools.

ASTM D86 distillation is carried out in a glass flask equipped with an electric

heater and a water cooler. The oil sample is heated in the flask and the resultant vapor is

condensed and collected in a receiver at laboratory pressure. The temperature versus

amount collected is recorded. Virtually no fi-actionation occurs in this distillation.

Equilibrium fla sh vaporization (EFV) is a test used to determine the vaporization

ratios of crude oils as a function of temperature. In this test, the crude oil is continuously

heated without separating the vapor from the remaining liquid (Nelson, 1958). The

temperature versus liquid amount is recorded. To obtain the EFV curve, a series o f runs

at different temperatures are earned out, and each run constitutes one point (of

temperature and percentage vaporized) on the flash curve.

103
 The equilibrium flash vaporization (EFV) distillation provides a relation between

system temperature and the percentage o f vapor generated. EFV distillation is a batch

process during which no vapor is separated. In principle, the heating process in EFV is

similar to that in the conventional distillation. In practice, the EFV curve is used to

estimate the vaporization ratio for the conventional design (Nelson, 1958).

Similarly, a close relation between ASTM D86 distillation and the stripping-type

distillation is apparent. Suppose a stripping-type tower has infinite number o f heaters

with a vapor withdraw line between adjacent heaters as well as zero pressure drop in the

tower, one would obtain exactly the same temperature vs. vaporization curve as ASTM

D86 distillation.

Now the ASTM D86 and EFV curves are used to answer two questions:

• For the same temperature limit, which distillation option produces more vapor, this is,

more distillates yield?

• To achieve the same vaporization ratio, which distillation option demands more heat?

The first question can also be rephrased as follows: to achieve the same

vaporization ratio, which distillation requires a higher temperature? This question is

important because we want to achieve the maximal vaporization at the allowable

maximal temperature. Although the answer to this is well-known, a review will highlight

the differences. Figure 4.4 shows a comparison o f the two curves.

The ASTM curve is located above the EFV curve. This means that at the same

temperature limit, ASTM distillation always produces less vapor than EFV distillation. In

this sense, the stripping-type design will likely produce less amount of distillates and

104
more residue under the same temperature limit. Whether other operating variables, such

as steam, could correct this problem remains to be answered.

500

Ü 400 A STM
lif EFV
K
g 300

u 200
Q.
s
UJ
I- 100

0 0.1 0.2 0.3 0.4 0.5 0.6

V A PO R FRACTION, W T

Figure 4.4 ASTM curve and EFV curve o f the Venezuela crude (see Appendix A)

The second question refers to which heating pattern is more energy-efficient.

From the viewpoint of energy efficiency, the stripping-type design reduces the heat

demand from the light components (naphtha and kerosene components), which are only

heated to low temperatures. Conversely, in the conventional design, light components are

heated to the temperature limit. The heating pattern for the stripping-type design (ASTM)

is, however, not advantageous for the heavy components (gas oil components). As

mentioned before, the distillation temperature for the ASTM to achieve the same

vaporization ratio is higher. This means that the heavy components vaporize at a higher

105
temperature than they would in the EFV distillation. Therefore, the heat demand for

heavy components is higher.

WTiether ASTM requires less energy to achieve the same vaporization ratio,

ignoring the temperature limit o f thermal cracking, will depend on the trade-off between

the heat saving for light components and the larger energy consumption for heavy

components. To obtain the ASTM heating curve, 9 heaters and 9 separators were used to

simulate the ASTM distillation. Figure 4.5 shows part o f the flow sheet used for this

simulation. The crude was heated at 760 mmHg step-by-step and vapors were withdrawn

in each step. The resulting curves for heat demand are shown in Figure 4.6.

VI
—►

Crude y ' T1 V2 Separator III

V3
Heater HI

Heater I Heater II L2 T3

Separator I Separator II L3 (to Heater IV)

►

Figure 4.5. Simulation of ASTM D86 distillation

Heaters 4 through 9 and separators 4 through 9 are not shown in this figure.

Figure 4.6 shows that for small vaporization ratios (less than 0.25), the two curves

coincide. For larger vaporization ratios, however, the heat demand for ASTM is always

106
higher than that for EFV distillation. To explain this, a comparison of temperatures and

heat demands needed for achieving 0.473 vaporization are listed in Table 4.1.

Because the vapor generated is not separated in EFV distillation, on average 24%

o f the crude in the form o f vapor is overheated from 100 °C to 307 °C. In ASTM

distillation, the overheating o f vapor is entirely avoided, as the vapor is withdrawn from

the system after it is generated. However, ASTM distillation has to further heat the liquid

(59% of the crude) from 307 °C to 380 °C. The above analysis means that the effort to

reduce the heat demand by avoiding overheating o f light components in the stripping-

type distillation may not be successful.

1000

Z — ASTM
800
— EFV
600
<
s
UJ 400
Q
< 200
UJ
z
0
0 0.2 0.4 0.6
VAPOR FRACTION, WT

Figure 4.6 Heat requirements for ASTM distillation and EFV distillation at 760

mmHg (lM3/hr,Venezuela crude, see Appendix A)

107
 Table 4.1. Comparison of temperature and heat demand

Temperature Crude vaporized Liquid remained Heat demand

°C Wt. fraction wt. MJ/hr
EFV 100 0.014 0.986 138.8
307 0.473 0.527 670
ASTM 307 0.345 0.655 582
380 0.473 0.527 720.3

4.4 Features of a stripping type design

As discussed in the previous section, the stripping-type design has to heat the

crude to a higher temperature in order to achieve the same amount o f vaporization as in

the conventional design. In addition, liglit components help vaporize heavy components

at lower temperatures. This is called the carrier effect and it was discussed in detail by Ji

and Bagajewicz (2001). Such light components are not present in the stripping-type

design.

The second feature is the different pattern o f product composition. In the

conventional design, all light components come from the bottom section of the tower and

go through the trays where side products are withdrawn. As a result, light components

appear in each side product withdraw line and have to be stripped off in the side strippers.

In industry, this is called controlling the flash point. In contrast, the light components in

the stripping-type design come from the top section and are withdrawn as soon as they

are vaporized, not reaching therefore the trays where diesel and gas oil are withdrawn.

Thus, the problem o f the presence of light components does not exist. However, heavy

108
components have a large chance of being present in these side products and raise their

end points.

The last feature is that the products are withdrawn in vapor phase instead o f liquid

phase. For the same composition of mixture, the dew point is higher than the boiling

point. So, by withdrawing a vapor one may obtain the heat at a higher temperature level.

This is advantageous from the point o f view o f energy efficiency. However, the vapor

phase withdrawals also bring the corrosion problem to the side condensers if water is

present.

4.5 Energy Targeting

The main column in Figure 4.1 contains 34 trays. A crude o f API 36 is used in the

simulation. The flowrate used is 5000 BBL/hr. The crude data were taken from a

previous chapter (Bagajewicz and Ji, 2001). The properties of the crude are listed in

Appendix B. Table 4.2 shows the product specifications.

Table 4.2 Stripping-type Column Specification

Specification
Naphtha D86 95% point, °C 182.2
Kerosene D86 95% point, °C 271.1
Diesel D86 95% point, °C 326.7
Tray 29 temperature, °F 360
(5-95) Gaps, °C
Kerosene -Naphtha 16.7
Diesel- Kerosene 0
Gas Oil- Diesel -13.9

109
 In simulating the stripping-type distillation, the feed temperature, the duties of

upper, middle and low heaters, the duties o f condensers are allowed to change in order to

meet the specifications. Comparisons are made with a conventional column operating at

maximum efficiency with the same specifications.

Ejfect o f feed temperature

Figure 4.7 is a heat demand-supply diagram for the stripping-type design. The

solid bold line stands for the crude heat demand curve, which is composed of the

preheating section and the upper, middle and lower side heaters. The heat supply includes

heat fi"om the main condenser, side condensers, products and the residue. The demand

curve is not continuous due to temperature gaps between adjacent heat sinks.

The discontinuity o f the heat demand curve is an important feature. It affects the

match between the heat demand and the heat supply. Consider the region between the

preheating curve and the upper heater. In this region, heat supply is available from side

condensers and the residue. However, there is no heat demand in this region. The only

way to make use o f this heat supply is to cascade the heat to regions at lower

temperatures. In the preheating region, a heat deficit is present (shaded region), but it is

much smaller than the heat supply aforementioned. Therefore, a large part of the heat

supply can not be utilized. Finally a small match between supply and demand exists at the

level o f UH.

110
 1.6

Pinch
MH
2
UH

i Heat deficit
0.8
a SC1
u Condenser Crude
SC3
LH

.4 SC2
Residue
SW
Res.
Kerosene
0
0 100 200 300 400
TEMPERATURE, C

Figure 4.7 Effect o f increasing the feed temperature

III
 The location o f the pinch point can be easily obtained from this diagram. It is the

lowest temperature at which the demand is larger than the supply after the shifting and

area matching has been performed (Figure 4.7). The heating utility is given by the

unmatched demand on the right, and the cooling utility is given by the extra supply on the

left.

The heating utility can be reduced by raising the temperature o f the feed. That is,

by shifting part o f the upper heater duty into the crude preheater. The solid curve in

Figure 4.8 represents the new heat demand curve for a higher feed temperature and the

dashed line corresponds to the basis case. Figure 4.9 shows the complete diagram for the

higher feed temperature.

1.6

Pinch MH
1.2
New demand UH
curve
0.8
u«

LH
0.4

0
0 100 200 300 400
TEMPERATURE, C

Figure 4.8 Effect o f increasing the feed temperature on heat demand

112
 1.6

MH

I- 0.8
SCI
UH
S Condenser Crude
SC3 LH

0 .4 SC2
Residue
SW
Res.
Kerosene

0 100 200 300 400

TEMPERATURE, C

Figure 4.9 The complete heat demand-supply diagram for the higher feed temperature

113
 Note that the location o f the pinch does not change at all, however the heat deficit

in the upper heater region decreases significantly. The demand of the middle and lower

heaters remains the same.

A comparison o f the strategy used above with that for the conventional design

reveals a few important issues. With a fixed heat demand curve (the crude curve) in the

conventional design, the strategy is to move the extra heat supply to higher temperature

regions, which was realized by redistributing the pump-around duties (Bagajewicz and Ji,

2001). In the heat demand curve in the stripping-type design can be modified by

redistributing heat among the preheater and the side heaters. Here the strategy is to move

some heat demand from the high temperature region (the upper heater region) to the low

temperature region (the preheater region).

However, there is a limit on the heat shift from the upper heater to the preheater.

Table 4.3 shows that the naphtha-kerosene gap decreases with the increase o f the feed

temperature. This is because the vapor arising fi’o m the upper heater decreases in amount

as a result o f a lower upper heater duty. In the section o f the column between the feed

tray and the upper heater, the vapor strips off light components from the descending

crude.

When the vapor flowrate is too low, the vapor cannot efficiently remove the light

components from the crude, and the crude entering the upper heater will carry a

significant amount o f light components. In the upper heater, these components vaporize

along with kerosene and enter the kerosene rectifier. As the rectifier has no way to

remove light components, all the light components are present in the kerosene product,

contributing to a lower 5% D86 boiling temperature.

114
 Table 4.3 Effect o f Feed Temperature in Stripping-type Design
237.8 °C 260 °C
Preheater duty, MW 60.0 72.4
Upper heater duty, MW 42.3 31.9
Naphtha yield, M^/hr 244.8 235.2
Kerosene yield, M^/hr 141.0 155.1
Naphtha-kerosene gap, °C 18.8 5.3

4.6 Comparison with the conventional design

The residue yield o f the stripping-type design is about 70% larger than that o f the

conventional design for the same temperature limit o f 360 °C. This yield cannot be

further reduced either by increasing steam injection or by adjusting the feeding

temperature and/or the duties o f the heaters. Because the residue yield has significant

effect in energy consumption, to make comparisons meaningful, the conventional design

was run at this large yield. As a result, the maximum temperature in the conventional

column drops to 324 °C. While the yields o f naphtha are almost the same, however, the

yield o f kerosene is significantly higher, and the yield o f diesel is lower in the stripping-

type design (Table 4.4). This is because in the conventional design, the light components,

which constitute the kerosene, ascend through the trays where the gas oil and the diesel

are withdrawn.

Some o f these light components are carried out with the diesel and the gas oil. In

the stripping-type design, the crude oil enters the column at the top. Therefore, the

carrying o f kerosene in the heavy products is ruled out. As expected, in the stripping-type

design the gap between kerosene and diesel is much higher than that in the conventional

design. Finally, the energy consumption is calculated: adding the minimum heating utility

115
o f the heat exchanger network and 70% o f the steam consumption, to account for cost

differences. The minimum utility is calculated using pinch analysis.

Table 4.4 Comparison o f Product Gaps, Yields and Energy Consumption

Conventional Stripping-type
Naphtha-Kerosene Gap (°C) 16.7 16.7
Kerosene-Diesel Gap (°C) 0.0 13.6
Diesel-Gas oil Gap (°C) -13.2 -13.9
Naphtha yield, M^/hr 244.7 242.1
Kerosene yield, M^/hr 137.8 157.4
Diesel yield, M^/hr 52.8 35.9
Gas oil )ield, M^/hr 43.9 35.0
Residue yield, M^/hr 316.4 324.5
Heating utility (furnaces, MW) 45.6 49.5
Steam consumption (MW) 19.1 16.4
Energy consumption (MW) 59.0 61.0

Simulations were performed for the stripping-type design using a higher

temperature limit (399 °C) to obtain the same yield o f residue as in the conventional

column (360 °C). It is at such higher temperatures, which are not recommended in

practice, when the stripping type design achieves around 6% less of total energy

consumption.

Investment cost

The above comparison shows the stripping-type design can not achieve the same

distillates yields as the conventional design for the same allowable heating temperature.

116
Since the crude is heated to a lower temperature, a less complex preheat train may be

needed. Thus the trade-off between operating and investment costs needs to be analyzed.

The relative costs of both designs are shown in Table 4.5. The main tower for the

stripping-type design is considered to be higher because the dirty crude goes through

most o f the trays. The dirty material reduces the tray efficiency and causes fouling

problems. Therefore, additional trays are needed to offset the lower efficiency and

especial designs are required to reduce fouling. Both requirements contribute to a higher

manufacturing and maintenance cost.

Table 4.5 Relative costs for both designs

Conventional Stripping-type Reason for

higher cost
Main tower lower higher Fouling
Furnace lower higher Heating more
cold streams
Heat exchangers similar similar
Side columns same same
Desalter same same
Main condenser same same
PA exchangers/side lower higher Corrosion
condensers
Total cost lower higher

The side condensers used in the stripping-type design are considered more

expensive than pump-around heat exchangers in the conventional design because the side

117
condensers have to deal with corrosion problem caused by the steam condensation. The

corrosion problem becomes more severe when processing high sulfur crudes.

The investment cost for the furnace in the stripping-type design is considered

higher because it handles four cold streams. The internal structure will be more complex

than that of the conventional design where only one cold stream exists. Besides the

energy consumption is higher.

The relative cost for all the other heat exchangers is not easy to estimate without

designing the heat exchanger network. However, a comparison can be made on the bases

of total number o f exchangers because it is the dominating factor in the total cost. The

total number o f exchangers is estimated using the heat demand-supply diagrams.

For the case o f the conventional colunm in the region above the desalter (Figure

4.10), there is one cold stream and 7 hot streams: P A l, PA2, PA3, residue, gas oil, diesel

and kerosene.

By using the (N-1) rule (Biegler et al, 1997), seven heat exchangers and a heater

are needed. The condenser is not included because the heat is in surplus and is cascaded

into the region below the desalter. In the region below the desalter, there is a large heat

surplus. Since the heat from the condenser is large enough, only one exchanger is needed.

In addition, the four distillates and the condenser have to be cooled to the final

temperatures, needing 5 coolers. Therefore the total number o f exchangers is 13.

For the stripping type design above 260 °C, 3 exchangers are needed for the crude

to extract heat from SC2, SC3 and the residue (Figure 4.11).

118
 1.4

Below desalter Below 260 C

260 C+ ATmi,
RES
PA1 CRUDE

^ 0 .6 COND

S 0.4 PA2 PA3

0.2
RESIDUE

100 200 300 400

TEM PERA TU RE, C
Products

Figure 4.10 Heat Demand-supply Diagram for Conventional Design

Products from the bottom up are: gas oil, diesel, kerosene and naphtha.

119
 UH

Below desalter Below 260 C

I
s 0.8
SC1
MH
COND
Ô 0-6 SC3 LH
SC2
RESIDUE
0.2
KEROSENE

100 200 300 400

DIESEL
GAS OIL TEMPERATURE, C

Figure 4.11 Heat Demand-supply Diagram for the Stripping-type Crude

Distillation

SW: saline water. SCI: side condenser 1. SC2: side condenser 2. SC3: side

condenser 3.

120
 The heat supply to the left o f UH has to be cascaded to lower temperatures for

utilization. Above the desalter and below 260 °C, there are 5 hot streams: the residue,

SC 1, kerosene, diesel and gas oil. Similar to the conventional design, the condenser is not

used. The number o f exchangers is 5 in addition to the heater. Below the desalter, one

exchanger is counted for heat exchange between the condenser and the crude. Four

coolers are required for cooling the products. The total number o f the exchangers besides

the heater is 13.

Both designs take the same number o f heat exchangers. Assume the number of

exchangers is the dominating factor in pricing, we expect the costs for the exchangers to

be close to each other. It is then concluded that the stripping-type design requires higher

investment cost than the conventional design.

4.7 Conclusions

From this study one can conclude that the stripping-type design for crude

fractionation cannot achieve low yields o f residue, and under the same yield of products,

it is not as efficient as previously reported. The difference may rely on different crudes

used as well as design procedures that utilize temperatures that are too high.

121
4.8 References

1. Bagajewicz M. and S. Ji. Rigorous Procedure for the Design o f Conventional

Atmospheric Crude Fractionation Units Part I: Targeting. Industrial and Engineering

Chemistry Research. 40 (2), pp. 617-626 (2001).

2. Biegler , L., Grossmann, I. And Westerberg, A., Systematic methods of chemical

process design, Prentice Hall, 1997.

3. Ji Shuncheng and Bagajewicz, M., Rigorous targeting procedure for the design of

cmde fractionation units with pre-flashing or pre-fractionation. Industrial and

Engineering Chemistry Research. Submitted.

4. Liebmann, K., Integrated crude oil distillation design, Ph.D. dissertation. University

of Manchester Institute o f Science & Technology, 1996.

5. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &

Chemical Engineering, 19, Supplement, SI 19, (1995)

6. Liebmann, K.; Dhole, V. R. and Jobson, M., Integration Design O f A Conventional

Crude Oil Distillation Tower Using Pinch Analysis. Inst, o f Chem. Engineers, 76(3),

part A, 335-347, (1998).

7. Glinos, K. and Malone, M. P., 1985, Minimum vapor flows in a distillation column

with a side-stream stripper. Ind. Eng. Chem. Proc. Des. Dev. 24(4); 1087-1090.

8. Nelson, W. L., Petroleum Refinery Engineering, 4* ed., McGraw-Hill, New York,

1958.

9. Sharma, R.; Jindal, A.; Mandawala, D.; Jana, S. K. Design/Retrofit targets of Pump-

around Refluxes for Better Energy Integration o f a Crude Distillation Column. Ind.

Eng. Chem. Res., 38, 2411, (1999).

122
10. Watkins, R.N., Petroleum Refinery Distillation, 2""^ edition. Gulf Publishers, 1979.

4.9 Appendix

Appendix A

The properties o f the light crude are presented. There are taken from Watkins

(1979). The crude is TIA JUANA light (Venezuela) having an API gravity o f 31.6.

Table 4A.1 TBP data for TIA JUANA light

Percent distillated Temperature, C

0 -3.0
5 63.5
10 101.7
30 221.8
50 336.9
70 462.9
90 680.4
95 787.2
100 894.0

Table 4A.2 Light ends for TIA JUANA light

Component Percent o f assay

C2 0.04876
C3 0.3762
I-C4 0.2774
N-C4 0.8908
Total 1.5932

123
Appendix B

The properties o f the crude (API 36) used in Table 4.3 and 4.4 are shown here.

There are taken from Bagajewicz and Ji (2001).

Table 4A.3 TBP Data

Percent distillated Temperature, °C
5 45
10 82
30 186
50 281
70 382
90 552

Table 4A.4 Light-ends Composition o f Crude

Component Percent o f assay
Ethane 0.13
Propane 0.78
Isobutane 0.49
n-Butane 1.36
Isopentane 1.05
n-Pentane 1.30
Total 5.11

Appendix C

Table C l lists the operating conditions of the designs reported by Liebmann

(1996). The designs were carried out using ASPEN plus release 8.5-6. Column 1 is the

124
stripping-type design. Column 2 is the conventional design. The conventional design was

simulated in this study using PRO II 5.11 and the results are listed in the last column.

Table 4A.5 Comparison o f Operating conditions in two types o f designs

Liebmann-D-2 Base-Liebmann*** Base-this study

Flash zone in, C 343 343
FZT,C 340 341.5
FZP,kpa 261 261
Heaters:
Heater-12 (UH) 195.2-248.5 C
Heater-23 (MH) 202.5-370 C
Heater-34 (LH) 340.6-370 C -
Flowrate, KG/S
LN 16.92 18.60 18.91
HN 13.5 12.22 12.21
LD 28.93 28.56 27.02
HD 12.81 13.93 22.10
RES 87.14 85.99 80.13
Steam, KG/S
HN-ST 0 0.21 0.49
LD-ST 6.5042 0.99 0.92
HD-ST 0 5.6MW* 0.72
RES-ST 5.1386 8.65 1.89
D86 5-95,C
HN-LN 14 16.1
LD-HN 19.5 16.0
HD-LD 5.5 7.7
*Reboiler ** Crude 100,000 BBL/day

125
 Chapter 5 Design of Crude Distillation plants with vacuum

units. I. Energy Targeting

5.1 Overview

In previous work, the targeting procedures for atmospheric distillation plants were

presented. These procedures were developed assuming that the atmospheric residue

leaves the plant as a product. In many cases, however, this residue is distilled in vacuum

towers to further extract more gas-oil. The addition of a vacuum tower has many

implications on energy consumption. The energy targeting o f three atmospheric

distillation-vacuum distillation combinations is performed in this article, revealing that

the addition o f a vacuum column favors the pre-flashing distillation arrangement, which

in turn is not as energy efficient as the conventional straight distillation arrangement

when the vacuum column is not utilized.

5.2 Introduction

A complete crude distillation process consists of atmospheric distillation and

vacuum distillation. Despite o f the importance o f crude distillation, studies on its optimal

design are scarce. One o f the important early publications is “Petroleum Refinery

Engineering” by Nelson (1936). This book discusses the crude distillation in detail. It

contains many charts and experimental data. The 4*'’ edition was published in 1958. In the

design procedure presented in this book, the amount o f products is estimated from the

crude distillation curve, then the number o f trays between side withdraws are picked.

Next, the temperature and pressure in each tray is calculated using empirical charts. Heat

126
balance is then carried out and the reflux is determined. The design procedure ends with

the design o f the preheat train.

Much later, another book on crude distillation was published by Watkins (1979).

This procedure provided in this book is similar to the one proposed by Nelson (1958). In

this design procedure, the distillation tower is designed first and followed by the design

of heat exchanger network. With this procedure, it is difficult to take into account the

interaction between the crude tower design and the heat exchanger network. Liebmann

(1998) proposed an integrated design procedure for the design o f the atmospheric tower.

The design procedure starts with a sequence o f simple columns that are generated by

decomposing the crude main tower. Next, reboilers and thermal coupling are introduced

in order to reduce utility consumption. The grand composite curve ( a tool used in Pinch

Technology) is used to assess the proposed modifications. After all the possible design

modifications have been explored, these columns are merged into a single complex

column.

Sharma et al (1999) proposed a method for calculating the maximum pump-

around heat removal. First, a practical minimum reflux ratio for each column section is

determined using Packie’s empirical diagram (Packie, 1941). Then, the heat removal in

the upper part o f the column is calculated using a heat balance. The upper part may start

from an arbitrary tray and end with the condenser. Following, the upper part is extended

tray by tray and heat surplus is calculated for each tray. The resultant heat surplus data

are used to construct a column grand composite curve. Finally, the maximum heat

removal for each section is determined using the column grand composite curve. A major

advantage o f this method is that the maximum heat removal can be estimated quickly

127
wiü-.out the need o f simulation. However, as Packie’s diagram is empirical and the effect

o f the stripping steam is not included, the heat removal calculated is not accurate. This

suggests a procedure based on rigorous simulations that can capture the relationships

between the column variables and the heat integration opportunities accurately.

Bagajewicz (1998) studied the flexibility aspects for the design o f atmospheric

columns using rigorous simulators. Bagajewicz and Ji (2001) proposed such a rigorous

targeting design procedure has been proposed for the design o f conventional crude

distillation units. This procedure aims at flnding the best scheme for a multipurpose crude

distillation unit that processes a variety o f crudes, which is an issue that had been

overlooked by previous design procedures. Heat demand-supply diagrams, instead o f

grand composite curves are used as a guide directing the search for optimal schemes. An

advantage o f heat demand-supply diagrams is that the role o f each stream, heater or

cooler in the total energy consumption is clearly shown, so the search of the best scheme

is straightforward. The trade o ff between different operating parameters is considered and

the decision is based on quantitative calculations instead of simple assumptions

(Bagajewicz and Soto, 2001).

Following the above procedure, the energy targets for a light crude and a heavy

crude were obtained. The targets were later used for the design of a multipurpose heat

exchanger network (Bagajewicz and Soto, 2001). The model for the multi-purpose heat

exchanger network takes advantage of the flexibility identified and is able to predict the

desalter temperature that needs to be used. A conjecture that a design for extreme crudes

would be able to accommodate the processing o f intermediate crude was posed and

partially confirmed.

128
 An important variant o f atmospheric distillation is pre-flashing or pre­

fractionation design. The pre-flashing and pre-fractionation designs are put in the same

category because both feature vapor separation before the furnace. Ji and Bagajewicz

(2001) presented a systematic procedure to obtain design targets for heat integration in

conventional crude fractionation units that use pre-fractionation columns or pre-flash

drums. It is shown that under the same high product yield conditions, pre-fractionation or

pre-flashing are not advantageous from the energy point of view. This is in great part due

to the loss o f the carrier effect that light components have in separating heavy gas-oil

fractions in the flash zone. The study also showed that the equivalent o f the carrier effect

o f lights could not be obtained by increasing steam injection.

Soto and Bagajewicz (2001) noted that extensive splitting o f crude streams has

several drawbacks. For example, not much energy efficiency is lost if the branching is

reduced and therefore the trade-off between heat utility and capital cost breaks even. The

model proposed Soto and Bagajewicz (2001) by uses binary variables to account for the

processing o f different crudes and count heat exchangers. The second part o f this paper

improves the model by allowing a match between a given pair to take place in different

intervals in different periods. Finally, none o f recent papers addressed the heat exchanger

network design for crude distillation with either the vacuum tower or with preflashing.

In this first part, the energy targeting o f three atmospheric distillation-vacuum

distillation combinations is performed. Alternatives for combined atmospheric distillation

and vacuum distillation are first studied. Next, targeting procedures for all alternatives

design are proposed. These procedures are used to find targets for the light crude. Finally,

the conventional-vacuum design is performed for a heavy crude.

129
5.3 Alternatives for combined atmospheric-vacuum distillation

Figure 5.1 shows the conventional design of an atmospheric unit. Crude is mixed

with water and heated in a heat exchanger network before entering a desalter where most

o f the water containing the salt is removed. The desalted crude enters another heat

exchanger network and receives heat from hot streams. Both heat exchanger networks

make use o f the vapors o f the main column condenser, the pump-around circuit streams,

and the products that need to be cooled. The preheated crude then enters the furnace,

where it is heated to about 340-370 °C. The partially vaporized crude is fed into the flash

zone o f the atmospheric column, where the vapor and liquid separate. The vapor includes

all the components that comprise the products, while the liquid is the residue with a small

amount o f components in the range o f gas oil. These components are removed from the

residue by steam stripping at the bottom of the column.

p . w ater
*■ n aphtha
PA l

PA 2 kerosene

PA 3 diesel
w ater D ESA LTER

C ru d e
HEN
gas oil
steam
so u r w a te r HEN FU R N A C E
R e sid u e lo Vac. T ow er

Figure 5.1 The conventional design

130
 In addition to the overhead condenser, there are several pump-around circuits

along the column, where liquid streams are withdrawn, cooled, and sent back to upper

trays. Such arrangement distributes the reflux throughout the column. It also has

beneficial effects for heat integration because the temperature o f these liquid draws is

higher than that of the condenser. Products are withdrawn in liquid state from different

trays and then stripped by steam in side strippers to remove light components. The

residue from the main tower is heated further in the vacuum furnace and sent to the

vacuum tower to extract vacuum gas oils.

The preflash design (Figure 5.2) is a variation o f the conventional design. It is

used for processing some liglit crudes.

condenser

naphtha
PAl ^

S te a m
PA2 ^ kerosene
Crude Oil
vapor
D esalter steam
PA3 ^
diesel

steam
sour w ater gas oil
Pre flash drum steam
Furnace AR to Vac. Tower

Figure 5.2 The pre flashing design

131
 In the preheat train, the crude is under pressure in order to suppress vaporization.

In some cases, however, the crude is so light that above certain temperature, the pressure

required to suppress vaporization is too high. The solution is to separate some light

components before heating the crude further in the preheat train. The light components

are sent to the main column directly.

A systematic procedure was presented (Ji and Bagajewicz, 2001) to obtain design

targets for heat integration in conventional crude fractionation units that use pre­

fractionation columns or pre-flash drums. It is shown that under the usual temperature

limits to prevent thermal cracking, the pre-flash design generates more atmospheric

residue and less gas oil. In a complete plant, the atmospheric gas oil loss can be picked up

by the vacuum tower. The price is that the vacuum tower has to handle a larger amount o f

feed stock and the vacuum jet steam consumption increases. Whether the trade-off is

beneficial is the question to be answered.

In the stripping-type design (Figure 5.3), the crude is heated to a relative low

temperature (about 150°C) and fed at the top o f the column. Because the crude

temperature is low, tlie vapor ratio o f the feed is small. The crude goes down the column

and is heated consecutively in three heaters (the upper heater, middle heater and lower

heater). Side products are withdrawn from the vapor phases, and rectified in side

rectifiers.

Liebmann and Dohle (1995) reported that for atmospheric distillation, the

stripping-type design could feature 5% less utility cost than the optimized conventional

design (Figure 5.1). However, the comparison is not appropriate because in the stripping-

type design, the crude oil was heated to a much higher temperature than in the

132
conventional design. The maximum allowable temperature is limited by the thermal

stability property o f the crude being processed and is found by lab testing. For the

Venezuela crude oil used in Liebmann’s paper, the allowable temperature is 343 °C

(Watkins, 1979). However, in the stripping-type design presented by Liebmaim (1996),

the crude was heated to 370 °C in both the middle heater and the lower heater (see

Appendix B). It can be expected that at this temperature, severe thermal cracking will

take place. Such operation is not allowed in practice. Because o f the above limitation of

their study, a re-evaluation o f the stripping-type design is necessary. Ji and Bagajewicz

(2001) took into account the temperature limit o f thermal cracking and made such

evaluation.

Preheater Water
> Naphtha
Desalted crude Kerosene

UH
^ Diesel

MH
► Gas oil

LH

Steam
>■ AR to Vac. Tower

Figure 5.3 The stripping-type design

133
 Similar to the preflash design, the loss o f atmospheric gas oil is picked up in the

vacuum column at the expense of extra vacuum jet steam consumption. In this paper, the

question of whether the total energy consumption is lower than in the conventional

design is evaluated.

5.4 Vacuum oil distillation

In this section features regarding vacuum distillation that are relevant for the

energy targeting procedure are highlighted.

The topped crude leaving the atmospheric tower still contains significant amount

o f valuable oils. As discussed above, these oils cannot be distillated at atmospheric

pressure because the temperature required would be so high that severe thermal cracking

takes place. Vacuum distillation is based on the principle that the boiling point of a

material decreases as the system pressure is reduced. In atmospheric distillation, the

whole crude cut points between distillates and the atmospheric residual are somewhere

from 370 to 430 °C. The cut point can be raised to 580 to 610 °C in vacuum distillation.

Thus, the general function of the vacuum tower is to remove the maximum possible

amount o f distillate from the charge stock while meeting product specifications (Watkins,

1979).

Figure 5.4 is a schematic diagram for production of light vacuum gas oil (LVGO)

and heavy vacuum gas oil (HVGO), but sometimes, depending on its properties, LVGO

is blended with other products like diesel. Both are typically used as feed to fluid

catalytic cracking units. The vacuum distillation consists o f the vacuum furnace, vacuum

tower and the vacuum producing system. The topped crude is heated up in the vacuum

134
furnace to about 400 °C. The temperature is controlled to be ju st below the temperature

of thermal decomposition. The heated oil partially vaporizes in the furnace and the vapor-

liquid mixture enters the flash zone of the vacuum tower. The vapor ascends in the

column and condenses in the HVGO pump-around section and the LVGO pump-around

section. Note that no strippers are used for the products because the initial boiling points

of LVGO and HVGO are not specified (Gary, 1994).

♦ To Ejectors

♦ VLGO

» VHGO
Topped Crude

Vacuum Furnace

Vacuum Residue

Figure 5.4 Vacuum Distillation (Watkins, 1979)

The HVGO section is the main heat removal zone. HVGO is withdrawn and

cooled down in the heat exchanger network and then a portion o f HVGO returns to the

top o f the packing. Although a single cut o f VGO (vacuum gas oil) is allowed in some

cases, drawing LVGO and HVGO separately is more beneficial from the point o f view o f

135
energy savings, because the resultant HVGO draw temperature is 90-120 °C higher than

the corresponding draw temperature o f a single VGO cut.

The LVGO section is the top zone in the tower where the LVGO is condensed

and separated from the non-condensables, cooled and a portion circulated to the top o f

the packing to remove heat.

The temperature o f the steam and non-condensable materials leaving the top o f

the vacuum tower is determined by setting a 10 to 25 °C approach to the minimum

practical cool oil temperature in the top pumparound circuit. The latter temperature is a

function o f the viscosity properties o f the oil processed. Usually a cooling oil temperature

of 65 to 95 °C is used. This allows an overhead temperature o f 90-135 °C.

The residue section serves two purposes: One is to remove relatively light

components, the other is to reduce coke formation. The former is achieved by steam

stripping using a steam to net residue ratio o f 4-6 Ib/bbl. The latter is realized by

circulating the partially cooled bottoms to quench the liquid to 365 °C.

The non-condensable gases and steam leave the tower from the top and are

ejected from the system by means of the steam jets, thus creating the required vacuum.

The non-condensables come from the following sources (Nelson, 1958):

1. Tail o f lower-boiling material.

2. Gases produced by cracking o f the feedstock.

3. Air dissolved in the feedstock and in the water used in generating steam.

4. Air leakage.

136
 The steam consumption consists o f two parts: process steam and vacuum jet

steam. The total steam consumption depends on the absolute pressure in the vapor line,

the amoimt of noncondensables, the vaporizer temperature and the temperature o f cooling

water. Based on the processing o f 1000 bpd o f a conventional Mid Continent topped

crude oil. Nelson showed that the most economical pressure for vacuum tower operation

is in the range from 30 to 100 mmHg. The steam consumption at 88 mmHg for different

flash zone temperatures is shown in Table 5.1.

Table 5.1 Vacuum distillation steam (process steam + vacuum jet steam)
consumption for 1000 bpd topped crude (Nelson, 1958)
Vapor line absolute Vaporizer Cooling water Steam consumption
pressure, mm Hg temperature, °C temperature, °C lb/day
88 360 27 58800
88 382 27 23280

The optimum performance depends on the absolute pressure in the flash zone and

the maximum non-cracking temperature at the vacuum heater outlet. Figure 5.5, taken

from Watkins (1979) shows the ratio o f gas oil distilled as a function of the flash zone

pressure and temperature.

The system pressure in this figure is the partial pressure o f the hydrocarbons

instead of the total pressure. In turn, the furnace outlet temperature depends on the

boiling range o f the feed, the fraction vaporized as well as on the feed coking

characteristics. Normally, the maximum flash zone temperatures in the vacuum tower

range from 413 to 427 °C (Watkins, 1979).

137
 The absolute pressure in the vacuum tower flash zone is controlled between 25

and 100 mmHg. The effective pressure may be reduced to about 10 mmHg by injecting

steam to the furnace and at the bottom of the vacuum tower. The amount o f stripping

steam used is a function o f the boiling range of the feedstock and the fraction vaporized,

and generally in the range o f 10 -50 Ib/bbl. The desired vacuum is maintained by the use

o f steam ejectors and barometric condensers. The number o f ejectors used is determined

by the vacuum needed. For a flash zone pressure o f 25 mmHg, three ejector stages are

required (Nelson, 1958).

120
0% 30% 50% 80%
100

20

300 400 500 600 700 800 900 1000

TEM PERATURE, F

Figure 5.5 Topped crude-vacuum region phase behavior (Watkins, 1979)

Vacuum distillation is widely used to produce catalytic cracking plant feed stocks

o f low carbon content. It is also used to produce lube oil fractions. In practice, the uses of

138
vacuum distillates and residue depend on the type o f crude oil feed, the type o f refinery

and its downstream processing capacities. The major specifications for the most common

products are (Watkins, 1979):

• Catalytic cracking feedstocks: Gas oils for catalytic cracking feedstocks require a

strict separation between distillate and residue. The amount o f carbon content in these

oils should be minimal while not sacrificing gas oil recovery. This is necessary to

reduce coke forming on the cracking catalyst. The metals content, particularly

vanadium and nickel, should be strictly limited because they are severe catalyst

poisons.

• Hydrotreaters or hydrocrackers Feedstocks: Feedstocks for hydrotreaters or

hydrocrackers can tolerate a slightly higher metal content because the hydrogenation

catalysts contain these metals. However, the amount of carbon and asphalt materials

should be minimal to prevent coke formation.

• Distillate fu el oils: Distillate fuel oils to be used directly are specified by API gravity,

viscosity, metal content and flash point. Atmospheric boiling ranges can be used to

define the separation.

• Vacuum residue: The properties of vacuum residue can be defined in several ways.

When distillate production is to be maximized, the amount of gas oil allowed

remaining in the bottoms stream must be minimized. The residue is blended into

residual fuels. In this operation, one can normally set the volume percent o f either the

whole crude or the topped crude that is to be yielded as vacuum residuum. This is in

effect equivalent to specifying a TBP cut point for vacuum distillates. The goal is to

139
 recover the maximum volume o f gas oil, which is free from contaminants by heavier

material.

5.5 Vacuum tower specifications

Table 5.2 shows typical specifications which are used in this study. This is a fuel-

type tower with two distillate products: light vacuum gas oil (LVGO) and heavy vacuum

gas oil (HVGO). In the refining industry, the specification for the LVGO is its flowrate,

and it is usually defined to be one third of the total distillate. In this study, however, the

same flowrate for LVGO cannot be specified in the three complete plants, because the

flowrate o f LVGO varies in a wide range. Therefore, D86 95% temperature is used to

specify LVGO for all the three designs.

Table 5.2 Vacuum tower specifications

Total trays 7
Flash zone pressure, psia 1.90
LVGO D86 95% temperature, °C 410*
Flash zone temperature, °C 382
Overflash ratio 0.02
Overhead temperature, °C 127
Bottom steam/vacuum residue, Ib/bbl 2-3 (2.74)
* Equivalent to 30 % vol percent o f total vacuum gas oil.

Products from the vacuum tower are cooled down to recover heat before being

sent for further processing. Naturally, one would like to recover as much heat as possible.

However, the final temperatures o f these products should not be too low, otherwise, they

become too viscous to pump. Table 5.3 shows the viscosity o f LVGO, HVGO and

140
vacuum residue at several temperatures. The final temperatures for LVGO, HVGO and

vacuum residue are specified at 60 °C, 82 °C and 177 °C, respectively.

Table 5.3 Vacuum Tower Product Temperatures

Product Temperature, °C Viscosity, CP
LVGO 104 3.9
60 14.2
HVGO 104 10.2
82 20.6
Vacuum Residue 104 1127
160 60.3
177 33.8

5.6 Targeting procedure

We now summarize the energy targeting technique for designing complete

distillation plants, a conventional-vacuum crude oil distillation plant. For the

conventional -vacuum design, the Watkins design method is used to obtain an initial

scheme without pump-around circuits. Then a heat demand-supply diagram is

constructed, and the direction o f heat shifting needed for maximum energy efficiency is

determined. This procedure may be repeated for other crudes.

Step 1: Begin with the lightest crude to be processed. As suggested previously, the

atmospheric tower does not have any pumparounds at this stage. The vacuum tower,

however, needs to have at least one top pumparound because it does not have a

condenser.

141
Step 2: The simulation is performed next. Usually simulation converges without

difficulty.

Step 3: Construct the heat demand-supply diagram.

Step 4: Transfer heat from the condenser of the atmospheric tower to the top atmospheric

pump-around.

Step 5: If product gaps (difference between 5% ASTM D86 temperature o f the heavier

product and the 95% ASTM D86 temperature of the lighter adjacent product) becomes

smaller than required, the stripping steam flowrate is increased to fix the gap. As long as

the steam added has a lower cost than the energy saved, one can continue shifting loads.

Otherwise, it is advisable to stop when a trade-off has been reached.

Step 6: If there is heat surplus from the pump-around circuit just added, transfer the heat

to the next pump-around circuit between draws in the same way as in step 4. If not, stop.

Step 7: For the vacuum tower, shift heat from the top pump-around to the low pump-

around. Check the total energy consumption and the product gap. The heat shifting

should stop when either the energy consumption does not decrease or the minimum

allowed gap bound is reached.

The energy consumption (E) is calculated using:

E = U + 0 .7 * '^ H ; (5.1)

where, U is the minimum heating utility of the system and H ' is the enthalpy of each

steam used in the column strippers. The weight factor o f 0.7 that multiply the steam

enthalpies is used because the low-pressure steam is cheaper than the fuel gas used as

142
heating utility. In turn, the minimum heating utility is assumed to be provided by a

furnace and is calculated using the well-known pinch calculation model.

The presence o f the vacuum distillation affects the optimal heat load distribution

in the atmospheric tower. With the variation o f heat distribution in the vacuum tower, the

pinch location may change. When Step 7 is completed, one should check the heat

demand-supply diagram and see whether the heat distribution in the atmospheric tower is

optimal. If not, one has to adjust the heat distribution and go to Step 5 again.

The pre-flash/pre-fractionation case: A special feature o f the preflash design is that the

heat demand for the crude can be adjusted by changing the temperature o f the preflash

drum. Because the total heat load in the atmospheric tower depends on the duty o f the

furnace, which decreases with an increasing preflash temperature, the preflash

temperature should be determined prior to distribution o f pumparound duties. Therefore,

in the case o f the preflash design, step 3 is modified as follows:

Step 3: Increase the temperature o f the preflash drum until the total energy consumption

starts to increase. During this procedure, product gaps are maintained by adjusting the

stripping steams.

The stripping-type atmospheric and vacuum case: Instead o f having several pump­

arounds as in the atmospheric tower of the conventional design, the atmospheric tower in

the stripping-type design has several heaters. With the knowledge that the conventional

design starts with no pumparounds, one can propose a stripping-type design to start with

only one heater. However, simulation shows this is not feasible because o f the

143
temperature limits. To overcome this problem, an initial heat distribution among the three

heaters is chosen. Thus, step 1 and step 2 are modified as follows:

Step 1: Begin with a stripping-type tower with an initial heat distribution. Set an initial

temperature for the feed.

Step 3: Increase the temperature o f the feed until the total energy consumption starts to

increase.

5.7 Results

It is shown in previous work (Ji and Bagajewicz, 2001) that both preflash and

stripping-type designs produce more residue than the conventional design. The higher

feed rate to the vacuum tower has two immediate effects. One is the requirement of a

higher capacity o f the vacuum tower and the other is a higher suction load to the vacuum

producing system. From the viewpoint o f energy consumption, both preflash design and

the stripping-type design have the advantage o f reducing the direct heat demand o f the

crude oil by avoiding overheating o f the light components. However, the corresponding

vacuum system consumes more steam. The trade-off both the two factors is here

investigated with the heat demand-supply diagram.

The feedstock used in this section is a light crude o f API 36. The property data for

this crude are shown in the Appendix A.

The conventional atmospheric-vacuum distillation design: Basically, there are two rules

to determine if a new pump-around is needed (Step 6). One is the existence o f heat

surplus in the pump-around last added and the other is that the addition o f a new pump­

around should reduce the total energy consumption. For the vacuum tower, the two

144
pump-arounds already exist at the beginning o f the procedure. The problem is then to

minimize the objective function without violating the product gap.

Table 5.4 shows the resulting energy consumption after the corresponding steps in

the targeting procedure are implemented. The energy consumption decreases constantly

with the addition of a new pump-around. The resulting heat demand-supply diagram is

shown in Figure 5.6.

Table 5.4. The variation o f the energy consumption with the addition of
each pump-around in the case o f the conventional-vacuum design
(Vacuum jet steam consumption not included)
- Energy consumption (MW)
No piunparound (Step 3) 97.29
With PAl (Step 4) 95.33
With PAl and PA2 (Step 6) 86.46
With PAl, PA2 and PA3 (Step 6) 76.94
With VPA2 optimized (Step 7) 75.13
ATmin = 22.2 ° C.

The installation of PAl or VPA2 only results in a small reduction of energy

consumption. This is because the temperature o f the heat provided by PAl is still low and

in the heat demand-supply diagram, most of this heat is in the region o f the condenser,

where heat supply is already in surplus. The reason why VPA2 results in a small decrease

is that the duty of the VPA2 is relatively small compared to the duties o f atmospheric

pump-arounds.

145
 1.2

Crude
PAl
0.8

I
0.6
Q. Cond VPA2
U < V4
s
^ V3
0.4
V4
Vacuum
PA1-
SW PA3 furnace
0.2 PA2 V4
gas oil
kero
0
100 200 300 400 500
diesel
TEM PERATURE, C

Figure 5.6 Heat demand-supply diagram for light crude

(atmospheric-vacuum plant)

146
 Although the heat loads provided by vacuum pump-arounds and products are not

as large as those o f the atmospheric counterparts, they affect the heat distribution in the

atmospheric pump-around circuits. With additional heat sources available in the

intermediate temperature region (PA2 region), it is beneficial for the complete design to

shift more heat from PA2 to PA3. Table 5.5 compares the heat distribution in the

atmospheric tower with the conventional design (Bagajewicz and Ji, 2001).

Table 5.5. Comparison o f pump-around loads in the atmospheric tower

(Units: MW)
Pump-around circuit Conventional design* Atmospheric -vacuum
design
PAl 22.28 ■ 23.07
PA2 33.71 12.87
PA3 8.79 22.69
* Taken from Bagajewicz and Ji (2001).

The complete plant has a lower duty in the middle pump-around (PA2) and a

higher duty in the lower pump-around (PA3). This can be explained as follows. In the

atmospheric conventional distillation only (see Figure 5.7), when the PA3 duty reaches

8.79 MW, the surplus in the PA2 region vanishes. Therefore, further heat shift to PA3

does not reduce the net heat demand and worsens the separation between diesel and gas

oil.

147
 1.4

.2

1 PINCH

g 0.8
CRUDE
dlQ.6 COND
0.4
PA2 PA3
0.2
RES
0
0 100 200 300 400

TEMPERATURE. °C

Figure 5.7 Atmospheric conventional distillation

148
 In the complete atmospheric-vacuum plant, however, vacuum products and pump­

arounds provide new heat sources in the PA2 region (Figure 5.6), which contribute to a

larger heat surplus in the PA2 region and allows more heat to be shifted from PA2 to

PA3. The duty of PA3 increases until the trade-off between the reduced energy

consumption and the increased steam consumption is not favorable.

Comparison: A comparison among the three complete plants is shown in Table

5.6. In this table from the left to right are the conventional design, the preflashing design

and the stripping-type design. The preflash design has the lowest energy consumption,

but its difference with the conventional is small, indicating that the conventional design is

competitive. The difference is because the atmospheric heater in the preflash design has a

lower duty than in the conventional design.

Figure 5.8 shows the heat demand-supply diagram for the preflash type crude

distillation. In this figure, the heat demand for the crude without preflashing is shown as

dashed line.

The Stripping-type design consumes 40% more energy than the conventional

design. Both the steam consumption and the minimum heating utility are higher than that

of the conventional design. The larger vacuum steam consumption is incurred by the

larger amount feed to the vacuum tower. The vacuum steam consumption of the

stripping-type design is 48.5% higher than the conventional design. The large increase in

the duty o f the atmospheric tower is due to the requirement o f heat at higher temperature

levels where less recoverable heat is available. The requirement for higher temperature

heat makes a large part o f the heat from the products and the pump-arounds unusable, and

more cooling water has to be used to remove the unusable heat.

149
 1.2
VPA1
V3
V2
PAl Pinch

Cond Crude
VPA1
d
VPA2
V4
0.4
V3
Vacuum
Furnace
0.2
Jsw
gas c
kero

100 200 300 400 500

diesel
TEMPERATURE, C

Figure 5.8 Heat demand-supply diagram for preflash type distillation (light crude)

150
 Table 5.6. Comparison o f the optimal designs o f three complete plants
Conventional Pre flashing Stripping-type
Steam enthalpy, MW
Atmospheric 15.74 13.08 16.38
Vacuum* 11.55 12.35 17.15
Total steam 27.28 25.43 33.53
Product Gap, °C
Naphtha-kerosene 16.6 16.7 18.8
Kerosene-diesel 0.0 0.0 0.7
Diesel-gas oil -4.0 -4.5 -25.8
LVGO-HVGO -30.4 -29.7 -30.2
Yield,M^/hr
Naphtha 244.39 244.92 244.81
Kerosene 144.76 145.15 141.03
Diesel 71.82 69.95 51.84
Gas oil 124.25 110.40 45.44
LVGO 22.92 34.92 115.09
HVGO 81.43 86.09 92.24
Vacuum residue 105.53 103.58 104.37
Vacuum overhead, kg/hr 171.95 219.04 92.08
HDBR**, MW 195.81 191.58
HU, MW 62.57 62.10 80.48
Energy, MW 81.67 79.90 114.01

* Steam consumption 23.28 lb per bbl feed.

Total heat demand before recovery

151
 2

.6
LH
u
.2
UH
HVGO MH
.8 LVGO VPA2
VRES /
SC3
CONDENSER
VPA VH
0.4

G
^Gas oil, 1 0 0/ / 200 SCI 300 400 500
SC2
Diesel VRES
TEMPERATURE. C

Figure 5.9 Stripping-type crude distillation (light crude)

152
 It is also seen that both the stripping-type design and the preflash design produce

less atmospheric gas oil, but the loss is picked up by the vacuum tower in terms of more

LVGO. The product gaps are similar among the three designs except for the diesel-gas oil

gap in the stripping-type design, where some diesel is not vaporized in the mid heater and

is carried out in the gas oil stream.

Figure 5.9 shows the heat demand-supply diagram for stripping-type crude

distillation. The addition o f the vacuum tower does not have significant effeci on the

topology of the heat demand-supply diagram. This is because the heat demand takes

place at high temperatures. The heat from vacuum products and vacuum pump-arounds

appears on the left o f these heaters, where the heat can only be used to preheat the crude.

While the heat from the vacuum section can be utilized to a large extent in both the

conventional design and the preflash design, it is almost o f no use in the stripping-type

design. This explains why the energy efficiency for the stripping-type design in a

complete plant is even worse than in the stripping-type atmospheric distillation plant.

5.8 Results for a heavy crude

As mentioned before, both the preflash design and the stripping-type design aim

at light crudes. In terms o f energy efficiency, the possible gain from both designs is that

the heat demand for the light components can be lowered (Ji and Bagajewicz, 2002a, b).

Because the amount o f light components in a heavy crude is low, it does not make sense

to use the preflash design or the stripping-type design for a heavy crude. Therefore, only

the conventional-vacuum design will be studied for the heavy crude. The heavy crude o f

20.0 API is used in this study. Property data o f this crude are shown in the appendix.

153
 The heating utility for a complete conventional plant processing a heavy crude is

51.35 MW. The heat demand-supply diagram and the operation variables for a scheme

with three pump-around circuits are shown in Figure 5.10. Compared with the

conventional design without vacuum distillation (Figure 5.11), the heat deficit in the low

to medium temperature range is improved but still prevails. The heat surplus in the PA2

region can be used to cover the heat deficit in the condenser region. Similar to the

atmospheric design, the total energy consumption is not sensitive to the atmospheric

tower heat distribution. This is an added flexibility for the heat exchanger network

design.

5.9 Conclusion

Rigorous targeting procedures have been developed for three types of complete

crude distillation plants. It has been found that the introduction o f vacuum tower changes

the topologies for both the conventional design and the preflash design, and thereby

changes the heat distribution among the pumparounds. In the stripping-type design,

however, the heat provided by the vacuum products cannot be utilized. The comparison

shows that energy consumption for the preflash-vacuum design is slightly smaller than

the conventional-vacuum design. Because the energy consumption difference is very

small, it should not be concluded that the preflash-vacuum design is better. The energy

targets obtained above are used in Part II to develop a heat exchanger network for a

complete distillation plant.

154
 .2

1
VPA2
0.8 VPA1 Crude

0.6

S 0.4
Cond
0.2 Residue
SW PA2
0
0 100 200 300 400 500
TEMPERATURE, 0

Figure 5.10 The conventional-vacuum crude distillation (heavy crude)

155
 0.8

0.6 CRUDE
PA3
üo
0.4

Q.
RES
ü
COND
* 0.2
PA1
SW PA2

0 100 200 300 400

TEMPERATURE, °C

Figure 5.11 The conventional crude distillation for the heavy crude

(without vacuum tower)

156
5.10 References

1. Bagajewicz M., On the design flexibility o f atmospheric crude fractionation units.

Chemical Engineering Communication, Vol. 166, 1998.

2. Bagajewicz M. and Ji S., Rigorous Procedure for the Design of Conventional

Atmospheric Crude Fractionation Units, Part I: Targeting, Industrial and

Engineering Chemistry Research, 40(2), 2001.

3. Edminister, W.C., Applied Hydrocarbon Thermodynamics, Houston, Gulf Publishing

Company, 1964.

4. Gary, James H., Handwerk, Glenn E., Petroleum Refining; technology and

economics, Marcel Dekker Inc., New York, 1994.

5. Golden, S., Prevent Pre-flash Drum Foaming, Hydrocarbon Processing, May 1997,

ppl41-153.

6. Harbert, W. D., Preflash saves energy in crude unit. Hydrocarbon Processing, July,

123-125, 1978.

7. Ji, Shuncheng and Bagajewicz, M., Rigorous targeting procedure for the design of

crude fractionation units with pre-flashing or pre-fractionation. Industrial and

Engineering Chemistry Research. 2002a, Submitted.

8. Ji, Shuncheng and Bagajewicz, M., On the energy efficiency of stripping-type crude

distillation. Industrial and Engineering Chemistry Research, 2002b, to be submitted.

9. Liebmann, K., Integrated crude oil distillation design, Ph.D. dissertation. University

of Manchester Institute o f Science & Technology, 1996.

10. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &

Chemical Engineering, 19, Supplement ,S119, 1995

157
11. Liebmann, K.; Dhole, V. R. and Jobson, M., Integration Design O f A Conventional

Crude Oil Distillation Tower Using Pinch Analysis. Institution o f Chemical

Engineers, 76(3), part A, 335-347, 1998.

12. Maxwell, J.B., Data Book on Hydrocarbon, Princeton, N.J., D. van Nostrand Co.,

1965.

13. Nelson, W. L., Petroleum Refinery Engineering, 4 ^ ed.. New York, 1958.

14. Packie, J.W. Distillation Equipment in the Oil Refining Industry. AIChE

Transactions. 1941, 37, 51.

15. Bagajewicz, M. and Soto, J., Rigorous Procedure for the Design of Conventional

Atmospheric Crude Fractionation Units. Part III; Trade-Off Between Complexity And

Energy Savings. Submitted. Industrial and Engineering Chemistry Research.

16. Watkins, R. N., Petroleum Refinery Distillation. G ulf Publishing Company, 1979.

5.11 Appendix

Appendix A

The crude oils used in this paper are summarized in Table 5A.1. The process rate

is 5000 bbl/hr. The TBP data and light ends composition are shown in Table 5A.2 and

Table 5A.3.

Table 5A.1 Feedstock Used for the Design

Crude Density Throughput
(kg/m^) (m^/hr)
Light Crude 845 (36.0 API) 795
Heavy Crude 934 (20.0 API) 795

158
 Table 5A.2 TBP Data (°C)
Vol. % Light Crude Intermediate Heavy Crude
Crude
5 45 94 133
10 82 131 237
30 186 265 344
50 281 380 482
70 382 506 640
90 552 670 N/A

Table 5A.3 Light-ends Composition o f Crude

Light crude Heavy crude
Compound Vol. % Vol. %
Propane 0.78 0.04
Isobutane 0.49 0.04
n-Butane 1.36 0.11
Isopentane 1.05 0.14
n-Pentane 1.30 0.16
Total 5.11 0.48

Appendix B

Table 5A.4 lists the operating conditions o f the designs reported by Liebmann

(1996). The designs w ere carried out using ASPEN plus release 8.5-6. Column 1 is the

stripping-type design. Colum n 2 is the conventional design.

159
Table 5A.4 Comparison o f Operating conditions in two types of designs

Liebmann-D-2 Base-Liebmann***
Flash zone in, C 343
FZT,C 340
FZP,kpa 261
Heaters:
Heater-12 (UH) 195.2-248.5 C
Heater-23 (MH) 202.5-370 C
Heater-34 (LH) 340.6-370 C
*Reboiler ** Crude 100,000 BBL/day

160
 Chapter 6 Design of Crude Distillation Plants with Vacuum

Units. II. HEN Design

6.1 Overview

In this chapter, the design o f the heat exchanger network is presented. A multi­

period heat exchanger network design model is proposed to handle two different crudes —

- a light crude and a heavy crude. This model considers the existence o f a preflash drum,

which is only used for the light crude period. Part of this multi-period model contains a

topological constraint through which all periods share the same heat exchange “matching

pattern” but not necessarily at the same temperature levels.

6.2 Introduction

In Chapter 5, rigorous targeting procedures have been developed for three types

of complete crude distillation plants. It was found that the introduction o f vacuum tower

changes the topologies for both the conventional design and the preflash design, and

thereby changes the heat distribution among the pumparounds. Comparisons show that

energy consumption for the preflash-vacuum design is slightly smaller than the

conventional-vacuum design. It was concluded that the stripping-type design cannot

compete with the conventional design in both the operating cost and the capital

investment. The energy targets obtained above are used in this part to develop a heat

exchanger network for a complete distillation plant.

161
 Bagajewicz and Soto (2001a) presented heat exchanger networks for atmospheric

distillation featuring maximum energy efficiency. The problem with such a design is that

it requires extensive splitting o f the crude stream. Such high splitting reduces the

controllability o f the preheat train. In another paper by the same authors (Bagajewicz and

Soto, 2001b), heat exchanger network for the atmospheric distillation using two, three

and four branches, respectively, were studied and showed the total cost is not sensitive to

the number o f branches. This is because the HEN with less number of matches includes

less exchanger shells and therefore less investment cost. It was concluded that when the

branching is reduced, the trade-off between heat utility and capital cost breaks even.

The model proposed in the above paper (Bagajewicz and Soto, 2001b) uses binary

variables to count heat exchangers and these binary variables are shared by both

operation periods. The sharing o f binary variables forces the the same matches for a

given pair o f streams in the same intervals. This paper extends the model to cases where a

match between a given pair can take place in different intervals in different periods. The

heat exchanger network also takes into account the flexibility o f using preflashing units

for light crude processing.

Although many heat exchanger network (HEN) design techniques are available,

their ability to address rigorous industrial problems with large number of streams and at

the same time controlling branching is limited. For example, Nielsen (1997) noted that

most HEN problems in the literature had less than 10 process streams. Papalexandri

(1998) and his coworkers noted that “industrial applicability o f the proposed methods (for

HEN) has been limited, as important features of industrial HEN can not been efficiently

account for”. For example, too complex heat exchanger networks with many branches

162
and complex piping should be avoided because o f process operability and reliability

concerns.

Besides the main objective o f the maximum heat utilization, the HEN should also

have the flexibility to accommodate varying operating conditions (Papalexandri, 1993).

In this regard, Bagajewicz (1998) and Bagajewicz and Ji(2001) showed that operating

conditions (process stream temperatures and flowrates) in a crude distillation plant vary

with the crude oil processed and the product specifications.

The problem o f multiperiod HEN has been addressed by several authors. Floudas

and Grossmarm proposed a procedure based on a superstructure representation that

embeds all possible structural options for different time periods (Floudas and Grossmarm,

1986, 1987). Papalexandri and Pistikopoulos (1993, 1994) proposed a unified

hyperstructure representation of mass and heat exchange alternatives to account for all

mass and heat integration possibilities.

The HEN design problem for a complete crude distillation plant discussed in

Chapter 5 has at least 18 process streams (15 hot and 3 cold). Compared with that of

chemical plants, the flexibility issue is much more challenging. Depending on the

difference o f crude oils to be processed, the stream flowrates and temperatures can vary

in such a wide range that a HEN for one crude oil is not even feasible for another crude

oil. With atmospheric distillation, Bagajewicz and Soto (2001) showed the optimal

topology for a light crude HEN departs substantially from that for a heavy crude. In fact,

the former is pinched while the latter is not.

This chapter addresses the design o f a heat exchanger network for a complete

crude distillation unit that uses a vacuum column for both crudes and a preflash unit for a

163
light crude. A new concept o f “match pattern” is presented and its mathematical

representation is given. This representation is based on the same interval transshipment

model as in Bagajewicz and Soto (2001) and extends this model to allow the crude to

match the same set of hot streams in different intervals and be considered the same

topology. A MILP model for such multi-period HEN design is proposed.

6.3 Energy Targets

Figure 6.1 shows the diagram for a complete crude distillation plant. Note that the

preflash drum is bypassed in the heavy crude period. The energy targets are used to

design the heat exchanger network are determined by the procedure proposed in Part I (Ji

and Bagajewicz, 2001).

condenser Vacuum tower

naphtha <r To Ejectors

P A l^
Preflash drum
steam
Crude Oil
vapor
PA2C kerosene X ‘ VLGO
steam

PA3C diesel ‘ VHGO

sour water
A m
g a so il
steam
ÏÏ
Desalter V Vacuum Residue
Furnace
Vacuum furnace
Heavy crude bypassing
preflash drum

Figure 6.1 Complete crude distillation plant

164
 The design procedure o f the distillation column starts by assuming a minimum

temperature difference for the heat integration or heat recovery approximation

temperature (HRAT). First, the Watkins design method is used to obtain an initial scheme

without pump-around circuits. Then a heat demand-supply diagram is constructed, and

the direction o f heat shifting needed for maximum energy efficiency is determined. This

procedure may be repeated for other crudes processed.

With this procedure and a HRAT value of 22.2 °C, the energy target for light

crude oil is obtained from Figures 2. The same procedure is applied for the heavy crude

case (Figure 6.3) and two sets of conditions can be obtained; product and pump around

flowrates, supply and target temperatures, ^ d energy consumption for the light and

heavy crude. The data that constitute the starting point of the design are shown in Tables

6.1 and 6.2.

It can be seen from the heat demand-supply diagrams that a heat exchanger

network featuring the minimum utility should be complex. For example, the pinch point

for the light crude (Figure 6.2) is 254 °C. Above the pinch, five hot streams are found in

the region between 254 °C and 310 °C. In order to completely use the heat from all the

five streams, the crude stream has to be split into five branches.

165
 2
V PA 1
V3
1 V2
PA1 Pinch

0.8

i Cond VPA1
Crude
0.6
a.
u VPA 2
s V4
0 .4
V3
Vacuum
Furnace
SW
2
gas oil
kero
0
100 200 300 400 500
diesel
TEMPERATURE, C

Figure 6.2 Heat demand-supply diagram for the light crude distillation

166
 VPA2
ü 0.8 VPA1 Crude

s 0.6
2 0.4
Cond
0.2 Residue
SW PA2

0 100 200 300 400 500

TEMPERATURE, C

Figure 6.3 The conventional-vacuum crude distillation (heavy crude)

167
 Table 6. I Stream data for light crude distillation with preflashing

Stream Description Ts. C Tt.C Q. MW MCP, MW/C

J1 crude before desalter 21.1 104.4 -38.83 -0.4658

J2 crude between desalter 162.8 383.4 -111.58 Vary

and preflash drum

J3 topped crude 348.9 413.4 -11.56 -0.1793

J4 crude between desalter 104.4 162.8 -26.61 -0.4560

and preflash drum

11 Saline water 104.4 43.3 5.73 0.0938

12 Atmospheric condenser 144.8 43.3 53.71 0.5290

13 Atmospheric PA1 176.7 104.4 17.59 0.2434

14 Atmospheric PA2 246.9 176.7 11.72 0.1670

15 Atmospheric PA3 306.1 232.2 20.52 0.2775

16 kerosene 185.0 43.3 10.52 0.0742

17 diesel 232.2 43.3 7.19 0.0380

18 Atmospheric gas oil 312.8 60.0 17.34 0.0686

19 Vacuum PA1 232.2 93.3 7.20 0.0518

110 Vacuum PA2 312.8 176.7 7.33 0.0538

111 LVGO 232.2 60.0 3.21 0.0186

112 HVGO 312.8 82.2 12.56 0.0545

113 Vacuum residue 371.1 176.7 15.20 0.0781

168
 Table 6. 2 Stream data for heavy crude distillation

Stream Description Ts, C Tt.C Q. MW MCP. MW /C

JI crude before desalter 21.1 137.8 -68.99 -0.5912

J2 crude between desalter 137.8 363.1 -123.51 Vary

and preflash drum

J3 topped crude 354.4 399.7 -22.06 -0.4872

11 Saline water 137.8 43.3 8.74 0.0925

12 Atmospheric condenser 119.1 43.3 12.83 0.1693

13 Atmospheric PA1 160.0 104.4 7.33 0.1319

14 Atmospheric PA2 256.5 176.7 10.26 0.1285

15 Atmospheric PA3 312.8 225.6 2.93 0.0336

16 kerosene 214.2 43.3 3.91 0.0214

17 diesel 288.6 43.3 10.00 0.0407

18 Atmospheric gas oil 278.8 60.0 3.23 0.0148

19 Vacuum PA1 225.6 93.3 19.69 0.1489

110 Vacuum PA2 312.8 176.7 7.33 0.0538

111 LVGO 225.6 60.0 10.93 0.0660

112 HVGO 312.8 82.2 10.29 0.0446

113 Vacuum residue 376.7 176.7 61.39 0.3069

169
6.4 Heat exchanger netw ork

The objective is to design a multi-period HEN with limited number of branches.

As mentioned before, the penalty for limited number o f branches is that the HEN requires

higher utilities. The requirements and assumptions are the following;

• Each cold stream can be split into at most 2 branches.

• Each heat exchanger can handle variable heat loads.

• Each cold stream exchanges heat with a fixed set o f hot streams in the same order in

all operation periods.

• HEN should have the ability to handle, different number o f process streams in

different period.

In the case o f light crude processing, a preflash drum is used to flash off light

ends to avoid two-phase flow in the preheat heat exchanger train. The preflash drum is

bypassed in the case o f heavy crude processing. Therefore, the light crude oil period has

one more cold stream.

The classical transshipment model (Papoulias and Grossmann, 1983) uses one

integer per pair of streams to determine the existence o f a match. This leads to many

streams transferring heat in the same interval and as a result, one cannot control splitting.

Bagajewicz and Soto (2001b) used integers as in the original transshipment model, but

they defined these integers in each interval. To count excahngers, they used a special set

o f constraints (Bagajewicz and Rodera, 1998).

170
 Although the model presented by Bagajewicz and Soto (2001) renders good

control o f branching, the model forces the two periods to share the same matches in the

same intervals. Figure 6.4 illustrates a cold stream C; matching with hot streams H, and

H] in two periods. The two topologies are apparently the same, that is, C\ received heat

first from and then from Hi. Only two heat exchangers are needed. However, the

above model can not handle this situation because in the second interval, the constraint of

single match in each interval is violated, even though the cold stream can cascade heat

up.

This paper uses two independent sets o f binary variables to count matches in each

period separately. A constraint is then applied which requires both periods share the same

topology. The topology is expressed in terms of matching patterns, which are defined in

the following section.

6.5 Stream matching pattern and its representation

Each cold stream is required to exchange heat with the same hot streams and in

the same order in all operation periods. An algorithm to represent the matching pattern

for each cold stream in each period is needed. Such representation should be unique for

each match pattern and be independent o f the intervals.

Bagajewicz and Rodera (1998) proposed to count heat exchangers using the

following constraint.

^,.y .r = m ax (y .y ^ - y .^ , 0) (6.1 )

In this constraint, y . j j represents a match between hot stream i and cold stream j

in interval T. This model count continuous matches as a single exchanger.

171
 HI mcp=l

H2 mcp=l

Cl m C p=l

Period I

© © 1 @
^---------<•------► m cp= l
HI

H2 i w m cp= l

Cl w # 1 m cp= l

Period II

Figure 6.4 The above figures show the same matching sequence between C l and hot

streams HI and H2.

172
 Now, consider three hot streams H |, Hz and H3 matching with cold stream J, as

shown in Figure 5. A parameter a, is associated to each hot stream.

HI H2 H3

Cl

0-0
Figure 6.5 A cold stream matching with three hot streams

Next another parameter PAj j is used to capture the information as to which hot

stream matches with cold stream j in interval T.

^^j.T - S ( 6 .2 )

If there is no match in an interval T, then PA^ j equals to zero. Next, define an

auxiliary variable PBj j as follows;

P B j j =mdûi{{PAjj C - (6.3)
*sr-i

where C is a number large enough to ensure that PBj j is non-negative. Further, define a

matching pattern identification number (MPIJ) for a specific sequence o f hot streams as

follows:

(6.4)

173
 Under certain choices of a, , MPIj is unique for each matching sequence. Table

6 . 3 shows the details corresponding to an example where J, matches with 3 hot streams

Hu Hz and H3.

Table 6. 3 Parameters representing a matching pattern

T1 T2 T3 T4 T5 T6
0 1 0 0 0 0

0 0 0 1 0 0

0 0 0 0 1 0

PAn.r 0 0 0

P P j\.T
0 0 0

M P Ij

C -l-a„y = C '-2 -a „ ^

The maximum operator in equation (3) can be represented in a linear form by

using the following equivalent constraints (Bagajewicz and Manousiothakis, 1992)

A,-hAC2>B
B>A,
B = M o x ( A j , A 2 ) <=> (6.5)
A2 + ( l - A ) n > B
B > A,

where X. is binary and Q is a sufficiently large number.

For matches between cold stream j and N hot streams, the possible number o f

matching patterns can be computed as follows;

174
Parent pattern and children patterns

Suppose a matching pattern for a cold stream involves K hot streams in sequence.

If a matching pattern is obtained by removing one or more hot streams without changing

the order o f the remaining hot streams, then the resulting pattern is called a child pattern.

Figure 6.6 shows an example o f a parent matching pattern and its six children patterns.

HI H2 H3
H2 H3

Cl
Cl
o - o — <> 0-0
(a) Parent pattern (b)

HI H3 HI H2

Cl Cl

0-0 0-0
(C) (d)

H2 HI H3
▲

Cl
o Cl
o Cl
o
(c) (0 (g )

Figure 6.6 Parent pattern and its children patterns

175
 Thus, for any given cold stream, the matching pattern for one period and has to be

either identical or a parent pattern o f the other periods.

Let MP(N, m) represent the number o f matching patterns for a cold stream to

match with m hot streams which are picked from a hot stream pool o f N hot streams.

Thus,

jV’
MP(N. m )= — (6.6)
m!

Let CP(K, i) be the number o f children patterns generated by removing i streams

from a parent pattern with K hot streams. CP(K, i) is given by:

CP{K,i) = (6.7)

Then the total number o f children patterns CP, is

(6 .8)
y

Conditions fo r a unique MPI

Consider a cold stream j matching sequentially with N hot streams I,,!?, L , • • In-i ,

In , from high temperature to low temperature. Therefore,

N -l
M/>/. (6.9)
1=1 j~ \

Different cases are now considered to infer the values of parameters a, to make

the sequence unique. Consider switching stream L and Im (k<m).. Then,

M P I) - M P I j = - ( N - k ) - a „ + ( N - k ) a, (6.10)

where M Pl) is the new pattern. After some manipulation, one obtains:

176
 MPIj - M PIj = { m - k ) {a^ -a „ ) ( 6 . 11)

Tlius, the condition for MPI j = M PI j is a* = a „ .

Consider now altering the order of three arbitrary hot streams Ik, I| and Im (k<l<m)

to a new order, say Im, Ik, Ii, keeping the positions of the other matches unchanged. Then

MPÇ-MPI=-{N-k) a^ +(N—k) a^^ + (N-r)-a, +(N-m)-a^

(6 .12)

Thus, the condition for MPI j = M PI j is:

_ { l —k) üj, + { m - l ) Q/
;
fn -k (6. 13)

In a similar fashion one can obtain the condition for M P Ij = M PI j when 4 hot

streams (1%, ly, 1%, I„) are exchanged is:

(6.14)
/+j + k +l

where i, j, k, I are positive integers that represent differences between the exchanged

positions. In view of the above, to make the value of MPIj unique, parameters a, should

satisfy the following conditions

• Be positive.

• Be different.

• No parameter can beequal to the average of any other numbers o f parameters.

Table 6. 4 lists parameters a- used in this study. It has been chosen to make a, as

hot stream number plus a decimal number. The purpose of picking such values is to

expedite data analysis.

177
 Table 6. 4 Parameter a, in this work

Stream Stream

11 1 .2 1 18 8.05
12 2.03 19 9.13
13 3.27 110 10.33
14 4.53 111 11.16
15 5.04 112 12.24
16 6.07 113 13.41
17 7.31 SI 20

6.6 Muiti-period model

In this paper, most o f the constraints by Bagajewicz and Soto (2001) are used.

However, instead of requiring the same matches interval by interval for different periods,

the matching pattern constraints are used. Following, the sets, variables and constraints

are listed.

Sets:

• Cold streams before splitting: JO

• Cold streams after splitting: J (including cooling water, which is placed at the last)

Subsets

JA=[J/, J 2 ] Ji and J 2 are two branches o f JO 1(crude before the desalter)

JB=[Jj, J 4 ] J 3 and J4 are two branches o f J02 (crude after the desalter)

JC=[Js, Jè[ J 5 and ie are two branches o f J03 (crude after the preflash drum)

178
• Hot streams including furnace: H ( furnace appears last)

HI = {Kerosene, Diesel, AGO, Residue, Naphtha, Sour-Water, Condenser,

PA l, PA2, PA3}

• Intervals T = (To, T i, ..., T n}

Variables:

Let 0(J) be split ratio o f branch J. For each period, the following equations hold:

Y ,d { ja ) = \ (6.15)
JA

Y ,e u b ) = \ (6.16)
JB

=1 (6.17)
JC

Q c ^ ,- e . Qco,^,, (6.18)

Q C j,,= e „ Q C O ,,„ (6.19)

Heat balance for hot streams:

R ^ j — R sx-\ ~ H U j ( 6 .20 )
j

^i.T (6 .21)
j

Heat balance for cold streams:

'EQu.r-QCj.r-O ( 6 .22 )

Upper bounds for heat exchange in interval T :

Qi.j.r ~ y ij.r (6.23)

Q ijx ~ ^ij.T ■y ijx - ® (6.24)

179
U ijj. 3i\d£. j j are positive constants. For simplicity, the same value can be used for all

ij , T.

A cold branch can only match with one hot stream (except heating utility) in each

interval.

(6.25)
i* S

T^i.j.T - y’i.J.T yi,J.T-\ T>T] (6.26)

In this constraint, jj. is a binary variable used to count heat exchangers. Note

that K, jj. can be one in a period even if the heat duty in that interval is zero. This stands

for a heat exchanger that is in use in one period and idle in another period.

Bound on the number o f heat exchangers;

E S N (6.27)
i.J.T

Equality o f matching patterns:

MPl'j = M Pl'j (6.28)

To reduce computational time, this problem is decomposed into two subproblems:

design above and design below the preflash drum temperature level. The objective

function for the subproblem above the preflash drum is to minimize the heating utility.

The number of exchangers N is initially given a large munber, say 50, and reduced step-

by-step until the utility consumption starts to increase. In addition, in this subproblem

residue heats in the last interval can be non-zero, allowing in this way to cascade down

180
heat to the other subproblem. As no heating utility is required below the preflash drum,

the number o f heat exchangers below the preflash drum is minimized.

6.7 Results and Discussion

Because the heavy crude bypasses the preflash drum, there is one less cold

streams in the heavy crude period than in the light crude period. In order to make the

number o f cold streams the same, the heavy crude is regarded as going through a dummy

preflash drum. This dummy preflash drum has neither temperature constraint nor vapor

effluent. By doing so, the heavy crude (after the desalter and before the furnace) is

divided into two segments: before and after the preflash drum (Figure 6.7).

Light crude

Heavy crude

0 100 200 300 400 500

TEM PERATURE, C

Figure 6.7 The starting and targeting temperatures o f colds streams in two periods

181
 Using the original intervals, heating utility would be required below the preflash

drum for the heavy crude. Table 6 . 5 shows the interval heat balances. Although the heat

balance for interval T , is positive, cold stream C4 has to split into at least 3 branches in

order to meet its heat demand.

Table 6 . 5 Heat balance for the heavy crude in the first interval below the preflash
drum
Interval T1 Heat cascaded from design above the
MW preflash drum, MW
Cl 0 .0 0 0 .0 0

C4 24.20 0 .0 0

11 0 .0 0 0 .0 0

12 0 .0 0 0 .0 0

13 0 .0 0 0 .0 0

14 6.28 0 .0 0

15 1.79 1.79
16 1.05 0 .0 0

17 4.56 2.57
18 1.51 0.79
19 7.31 0.03
110 6.03 3.40
111 3.26 0.03
112 3.56 1.38
113 15.01 0 .0 0

Interval heat from hot 26.16

streams

182
 Table 6. 6 compares the targeted furnace duty and the actual duty. The duties for

both the light crude and the heavy crude increase as expected. There are 8 exchangers

(including the furnace) above the preflash drum and 17 exchangers below, totaling 31

exchangers. In the light crude period, hot streams H I, H6, H7, H8 and HI 1 are not used.

Some o f these streams are hot enough to heat the crude above the pinch.

Table 6. 5 Comparison o f targeted furnace duty and actual furnace duty

Light crude Heavy crude

Targeted furnace duty, MW 68.02 57.26
Actual furnace duty, MW 71.55 57.94
HRAT/EMAT=33.3 °C/22.2 °C

However, they are not selected before in the same temperature interval there are

other hot streams with larger heat capacities. As the cold streams are limited to split into

two branches, the model picks two hot streams with the largest heat capacities. These

streams cannot be used below the desalter because the heat from the condenser is enough

to heat the crude. The situation changes with the heavy crude, where the aforementioned

hot streams are used to heat the crude below the desalter, because the condenser heat is

not enough (see Figure 6.3). The different heat supply scenarios below the desalter

increase the complexity o f the heat exchanger network. As shown in Figure 6.10, nine

exchangers are required to heat the heavy cmde, while only one is used for the light crude

(Figure 6.10). The situation is similar to that found by Bagajewicz and Jose (2001) with

the atmospheric distillation.

183
6.8 Conclusion

The paper addresses the design o f a heat exchanger network for complete crude

distillation plants. A multi-period heat exchanger network design model was proposed to

handle two radically different crudes. In order to reduce the complexity o f the heat

exchanger network, an assumption is made that crude streams can be split to no more

than two branches. This model also takes into account the flexibility o f using preflash

drum for the light crude only. Part o f this multi-period model contains a topological

constraint through which all periods share the same heat exchange “matching pattern” but

not necessarily at the same temperature levels.

10.33
247 C 185 0
H4
306 C 16.45
o 247 0
HS
313C
o 8.77 185 0
H8
313 C
o 306 0
HIO
4.29
313 C 234 0
H12
o
H13
371 C 9.70

o 247 0

61.70
256 0 207 0 194 0
383 C 257 C

CXX) 0 .6 6 163 0
02
o
270 C 244 0 0.34

11.56 o o
4130 349 0
03

Figure 6.8 Heat exchanger network above the pieflash drum for the light crude

184
 3.98
257 C 226 C
H4
313C 1.14 279 C
H5
279 C 279 C
H8
313 C 1.30 289 C
HIO
2.51
313 C 257 C
H12
377 C 46.40 226 C
H13

36.17
2 60C 252C 233 C
363 C 270 C 0.245 203 C

0% H D C2

316C 0.755
22.06
400 C 354 C

Not used

Figure 6.9 Heat exchanger network above the preflash drum for the heavy

crude

185
 8.66
145 C
H2 43 C
7.78 9.96
177 C 145 C
H3 !0 4 C

232 C
12
H9 93 C
6.56
306 C 184C 0.38
HIO 177 C
3.80
1.67
247 C 198 C
H13 177 C

38.83
I0 4 C
21 C
Cl

163 C 0.36 1 04C

C4
163 C
150 C 0.64

Figure 6.10 Heat exchanger network below the preflash drum for the light

crude

( Coolers for H I, H6, H7, H8, HI 1 and H12 are not shown, idle exchangers on

stream C l can be found in Figure 6.9)

186
 8.77
138 C 43 C

12.83
119 C 43 C
H2
7.33
160 C 104C
H3
6.34
177C
226 C
H4
8.73
1.26
289 C 258 C
H7

226 C 93 C
H9
7.76
226 c 108 C
H Il
5.31

H12
257 C 139 C

i;
56 C 0.42
138 C 117C 86 C
Cl

21 C

0.58
138 C 134 C 58 C

Figure 6.11 Heat exchanger network for heavy crude (C l) below the desalter

H3
9.76 9.98
226 C 160C 93 Ç
H9
6.03
289 C 177 C
HIO
15.04
226 C 177C
H13

203 C
0 56 ,38 c
C4
203 C
0.44

Figure 6.12 Heat exchanger network for heavy crude (C4) below the preflash

drum

187
6.9 N om enclature

£,■j T = positive number used in the logical inequality

0(J) = split ratio o f branch J

CP(K, i) = the number of child patterns generated by removing i streams

E = total energy consumption

H- = enthalpy o f steam i

HUj- = heat from the hot utility in interval T

I = hot stream

J =cold stream

MP(N, m) = the number of matching patterns

Ki jj. = variable used to count heat exchangers

PAj j. = parameter representing a match between a cold stream j and hot stream i in

interval T

PBj j = auxiliary variable for representing a heat exchange match pattern

Q ijx ~ heat transfer between hot stream i and cold stream j in interval T

QH^ j = heat available from hot stream i in interval T

QCj j = heat demand for cold stream j in interval T

Qsj.t~ heat transfer between hot utility S and cold stream j in interval T

Rsx —residue heat of hot utility in interval T

/?, ;•= residue heat of hot stream i in interval T

T —interval

188
U = minimum heating utility

~ positive number used in the logical inequality

y, y r - binary variable for heat transfer bet^veen stream i and cold stream j in interval T

6.10 References

1. Andrecovich, M., and Westerberg, A., A Simple Synthesis Method Based on Utility

Boimding for Heat-integrated Distillation Sequences. AIChE Journal, 31(3), 363-375,

(1985).

2. Bagajewicz M. and S. Ji. Rigorous Procedure for the Design of Conventional

Atmospheric Crude Fractionation Units Part I: Targeting. Industrial and Engineering

Chemistry Research. 40 (2), pp. 617-626 (2001).

3. Bagajewicz M. and J. Soto., Rigorous Procedure for the Design of Conventional

Atmospheric Crude Fractionation Units Part II: Heat Exchanger Networks. Industrial

and Engineering Chemistry Research. 40 (2), pp. 627-634 (2001a).

4. Bagajewicz M. and Soto. J., Rigorous Procedure for the Design of Conventional

Atmospheric Crude Fractionation Units. Part III: Trade-Off between Complexity and

Energy Savings. Submitted. Industrial and Engineering Chemistry Research.(2001b)

5. Bagajewicz, M., On the Design Flexibility of Crude Atmospheric Plants. Chemical

Engineering Communications, 166, 111-136,(1998).

6. Bagajewicz, M. and Manousiouthakis, V. Mass/Heat-Exchange Network

Representation o f Distillation Networks. AIChE Journal, 38(11), 1769 (1992).

189
7. Bagajewicz M. and H. Rodera. A New MILP M odel For Heat/Mass Exchanger

Networks Featuring Minimum Number O f Units. AIChE Annual Meeting, Miami

(1998).

8. Floudas, C. and Grossmann, I., Automatic generation o f multiperiod heat exchanger

network configurations. Computers and Chemical Engineering, 11(2), 123-142

(1987).

9. Floudas, C. and Grossmann, I., Synthesis o f flexible heat exchanger networks for

multiperiod operation. Computers and Chemical Engineering, 10 (1), 153-168(1986)

10. Golden, S., Prevent Pre-flash Drum Foaming, Hydrocarbon Processing, May 1997,

ppl41-153.

11. Gundersen T. and I. E. Grossmann, Improved Optimization Strategies for Automated

Heat Exchanger Network Synthesis Through Physical Insights. Comput. Chem.

Engng. 14, 9, pp. 925-944 (1990).

12. Ji, S. and Bagajewicz M., Rigorous targeting procedure for the design o f crude

fractionation units with pre-flashing or pre-fi-actionation, lECR, submitted.

13. Liebmann, K.; Dhole, V. R. Integrated Crude Distillation Design. Comput. Chem.

Eng. 1995,19, SI 19.

14. Liebmann, K.; Dhole, V. R.; Jobson, M. Integration Design O f A Conventional Crude

Oil Distillation Tower Using Pinch Analysis. Institution o f Chemical Engineers,

1998, 76(3), part A, 335.

15. Nielsen, J., et al. Retrofit and optimization o f industrial heat exchanger networks: a

complete benchmark problem, Computers and Chemical Engineering, 21, S469-474

(1997).

190
16. Papalexandri K, et al. Heat integration aspects in a crude preheat refining section.

Computers and Chemical Engineering, 22, S I41-148, 1998.

17. Papalexandri, K. and Pistikopoulos, E., a multiperiod MINLP model for improving

the flexibility o f heat exchanger networks. Computers and Chemical Engineering, 17,

8111-116(1993).

18. Papalexandri, K. and Pistikopoulos, E., a multiperiod MINLP model for the synthesis

of flexible heat and mass exchange networks. Computers and Chemical Engineering,

18(11-12), 1125-1139, 1994.

19. Papoulias S. and I. E. Grossmann. A Structural Optimization Approach in Process

Synthesis-II: Heat Recovery Networks. Comput. Chem. Engng. 7, 707 (1983).

20. Sharma, R.; Jindal, A.; Mandawala, D.; Jana, S. K. Design/Retrofit targets o f Pump-

around Refluxes for Better Energy Integration o f a Crude Distillation Column. Ind.

Eng. Chem. Res. 1999, 38,2411.

21. Soto, J., Multipurpose heat exchanger network for maximum energy efficiency of

crude fractionation units. Master Thesis, University o f Oklahoma, 2000.

22. Stichlmair, J.G, and Fair J. R, Distillation Principles and Practices, Wiley-VCH, New

York, 1998.p76.

23. Terranova, B., and Westerberg, A., Temperature-Heat Diagrams for Complex

Columns. 1. Intercooled/Interheated Distillation Columns. Ind. Eng. Chem. Res. 28,

1374-1379, (1989).

191
 Chapter 7 New Flowsheets for Crude Distillation

7.1 Introduction

In previous chapters, systematic methods were presented and applied to crude

distillation design. Rigorous procedures were proposed for the design of both the

atmospheric plant and the complete plant. The procedures take into account the flexibility

o f processing different crudes year around. In addition, the role o f light components in

crude fractionation known as the carrier effect or the carrier design was first given a

rigorous definition. In view of this, two typical design alternatives, the preflashing design

and the stripping-type design were analyzed focusing on energy efficiency and compared

to the conventional design.

In addressing the aforementioned methods, a multi-period heat exchanger network

design model (Chapter 6) was proposed to handle two radically different crudes — a light

crude and a heavy crude. Part o f this multi-period model contains a topological constraint

through which all periods share the same heat exchange “matching pattern” but not

necessarily at the same temperature levels.

This chapter explores possible new flowsheets for future study. Two tools are

used: the heat demand-supply diagram and distillation sequencing.

7.2 Use of the heat demand-supply diagram

The heat demand-supply diagram provides a straightforward view of the energy

recovery scenario. Figure 7.1 shows the diagram for the conventional atmospheric crude

192
distillation. Efforts on réduction o f energy consumption should be made in the following

two directions:

• Increase the quantity or quality o f the heat supply

• Decrease the quantity o r quality o f the heat demand

0.9 PA1
0.8 Crude
0.7

I
CL
0.6
0.5
Cond

PA1
U 0.4
s PA3
0.3
0.2 PA2
<ero Res
0.1
GO
0 100 200 300 400 500
TEMPERATURE, C

Figure 7.1 Atmospheric Distillation (light crude oil, ATmin - 22.2 °C)

The properties o f the crude can be found in Tables 2.1, 2.2 and 2.3.

Cond: condnser. Kero: kerosene. GO: gas oil. Res: residue. PA: pump-around.

The first case was discussed in Chapter 3. In the second case, one wants to vary

the heating pattern o f the crude, as in the case of the prefash design. In trying to obtain an

alternative, focus is shifted to the utilization of heat from the condenser. The major hurdle

preventing the use of the condenser heat is its low temperature level. The most

193
convenient way to overcome this hurdle is to shift heat from condenser to the pump-

arounds (Bagajewicz, 1998, Bagajewicz and Ji, 2001). However, this effort is limited by

material and heat balance. Specifically, as heat is shifted, the liquid flowrate in the tray

above the pump-around is reduced. When this reaches zero, the maximum transferable

heat is also reached. Now ways to break this limit are explored.

One way to shift the condenser heat to higher temperatures is by using a heat

pump (Figure 7.2).

C ru d e
HE

High Temperature
H eat out (Q ^)
(T h)

H eat W ork in (W-^)

Pum p

Low Temperature
H eat in (Q^)
(JO

C ondenser

Figure 7.2 Reducing heat consumption using heat pump

This alternative does not alter the operating conditions of the column, but is

expensive. Table 7. 1 shows the relative cost for shifting heat from the condenser and

194
PAl to the targeted region (Figure 7.3). Note that the heat is transferred through the

pinch.

P in ch

Crude
Ü 0.7
Cond
- 0.5
O 0.4

0 100 200 300 400 500

TEM PERATURE, C

Figure 7.3 Shifting heat to a high temperature region using heat pump

The properties of the crude can be found in Tables 2.1, 2.2 and 2.3.

The pinch location is shown by a dash line

In order to estimate the cost, the electricity consumption needs to be estimated.

The theoretical minimal work (ideal work) needed for transferring the heat can be

computed using equation (1):

195
 (7.1)
^H

where fV-^ is the amount of ideal work consumed

7^ is the source temperature at which the heat pump extracts heat.

T;, is the target temperature at which the heat pump discharges heat.

Qff is the amount o f heat the heat pump provides at 7^ .

The actual electricity consumption can be computed by equation (2).

(7.2)

where

ir is the actual (real) work consumption.

rj is the heat pump efficiency.

In Figure 7.1, it is intended to transfer 13.25 MW heat from the condenser

through the pinch. The temperature range o f the heat available is 144.1-170.6 °C (417.3-

443.7 °K) and the targeted temperature range is 201.5-254.4 °C (474.7-527.6 °K).

Assume two heat pumps are used, each transferring 6.625 MW heat from the condenser.

The first heat pump operates between 144.1 C and 227.9 C, and the second heat pump

operates between 157.4 C and 254.4 C. The heat pumps have an efficiency of 60% (Table

7.1). Based on the prices provided by Douglas (1988, fuel S4.00/MMbtu, electricity

$0.04/kwhr), the annual saving in operating cost is $245,406 per year.

196
 Table 7. 1 Relative operation cost for shifting heat to the targeted region

Heat pump A Heat pump B

T„.K 501.1 527.6
T l, K (condenser) 417.3 430.5
Ww/Qn 0.17 0.18
Heat pump efficiency 0.60 0.60
W /pH 0.28 0.31
Q l, MW 6.625 6.625
Q h, MW 7.96 8.12
Wr, MW 2.22 2.49
Annual savings in fuel*, S 859733 877357
Annual cost for the electricity, S 702785 788899
Net annual savings, S 156947 88458
* 330 days/year
** fuel S4.00/MMbtu, electricity S0.04/kwhr

7.3 Distillation sequencing

Distillation sequencing is another useful tool for developing new design schemes.

The problem o f crude distillation can be viewed as to be similar to separating 5

components. The conventional design is virtually an indirect sequence, as it is shown in

Figure 7.4. The bottom parts o f columns I, 2 and 3 correspond to the side strippers in the

conventional design (Figure 7.5). The stripping-type design is a direct sequence (Figures

7.6 and 7.7).

197
 I I Condenser

1-------------- ► Naphtha

^----- Steam
----------- ► Kerosene

- Steam
Diesel

Steam
Crude ^ Gas oil

Residue

Figure 7.4 Indirect sequence

water
<------- '------- ►na
naphtha

PAl

steam Stripper 3
PA2
c : kerosene

steam Stripper 2
PA3
c : diesel

C rude steam Stripper 1

% gas oil
Flash zone
steam

Residue to vacuum Tower

Figure 7.5 Conventional design

198
 Naphtha
Crude

-► Kerosene

J
ÜH Diesel

^-0 -
MH Gas oil

^— 0 -
LH ■* ----- Steam

Residue

Figure 7.6 Direct sequence

Water
Prch cater
Naphtha
Crude oil ^ Kerosene

Rectifier
UH
Diesel

Rectifier 2
MH

Gas oi

Rectifier 3
LH

Steam

Residue to vacuum tower

Figure 7.7 Stripping-type design

199
 Preflashing and Prefractionation designs are other sequences (Figure 7.8, 9). Note

that in the preflash design the sloppy separation o f the first flash produces a top vapor

that needs further separation. In this regard, this scheme departs from the traditional sharp

sequence proposed as framework for the analysis. The complete list of possible

sequences is depicted in Figure 7.10.

» Kerosene
PA2
* Naphtha

Crude Diesel
PA3

Gas oil

Residue

Figure 7.8 Prefractionation design

♦ N a p h th a

PAl

Kerosene

Crude
Diesel

♦ Gas oil

Residue

Figure 7.9 Pre flashing design

200
 N K
K D
-► — Stripping-type design
D G R
G R
R
K
N
D
G
K K
N D R
D
K G G
D K
R Prefractionation design
G D

N
K
D N
Conventional design
K

Figure 7.10 Distillation sequences for crude separation

N-naphtha, K-kerosene, D-diesel. G- gas oil, R-residue.

In screening the potential sequence, one has to take account of the following

constraints

1. The maximal allowable temperature

2. The allowable residue yield

201
3. Non-sharp split between distillates

These constraints can be used to eliminate some sequence. For example, the first

and the second constraints exclude the stripping-type design. A previous study (Ji and

Bagajewicz, 2001) shows that early separation of stripping components increases the

yield o f the residue, which is not a desired effect. The stripping components include all

compounds in naphtha and kerosene as well as some components in diesel. Therefore, the

residue has to be separated first in order to maintain the stripping effect as in the bottom

branch which eventually leads to the conventional design but renders others too. The

remaining sequences are meaningful and deserve to be explored.

7.4 Vacuum distillation

Vacuum distillation is used to vaporize those components that cannot be flashed

using atmospheric pressure. The current crude vacuum distillation only uses one column

operating at about 10 mmHg. Because the vacuum jet steam consumption increases with

the decrease o f the distillation pressure, it is natural to think about an alternative of using

two vacuum towers, one operating at low vacuum, say 100 mmHg, and the other

operating at 10 mmHg (Figure 7.11). If a significant amount o f vacuum gas oils can be at

the low-vacuum tower, the feed to the second vacuum tower can be reduced and

accordingly the total vacuum jet steam will be reduced. The disadvantage is the extra

investment cost o f the low-vacuum tower. Hopefully, the cost is not high. As the vapor

volume is inversely proportional to the operating pressure, the diameter of the low-

vacuum tower will be only one tenth of the 10 mmHg tower.

202
 To Ejectors To Ejectors

Heavy
X
Topped
crude
»
n
(-—\
diesel
VLGO-I a X
-► VLGO-II
-► VHGO

Vacuum Vacuum Vacuum

Heater-1 Heater-2 residue
Vacuum Tower-1 Vacuum Tower-2
(100 mmHg) (10 mmHg)

Figure 7.11 Vacuum distillation with two towers

Other separation technologies, such as membrane separation, could be

incorporated with distillation. This is part o f the future work.

203
7.5 References

1. Bagajewicz, M. On the Design Flexibility o f Crude Atmospheric Plants. Chemical

Engineering Communications. 1998, 166, 111.

2. Bagajewicz M. and S. Ji. Rigorous Procedure for the Design o f Conventional

Atmospheric Crude Fractionation Units Part I: Targeting. Industrial and Engineering

Chemistry Research. 40 (2), pp. 617-626 (2001).

3. Douglas, J. M., Conceptual design o f chemical processes, p569, McGraw-Hill, 1988.

4. Ji Shuncheng and Bagajewicz, M., Rigorous targeting procedure for the design of

crude fractionation units with pre-flashing or pre-fractionation, Industrial and

Engineering Chemistry Research. Submitted.

5. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &

Chemical Engineering, 19, Supplement, SI 19, (1995)

6. Stichlmair, J.G, and Fair J. R, Distillation Principles and Practices, Wiley-VCH,

New York, 1998.

204